Datasets:

DanielPFlorian

/

Transformers-Github-Issues

like 0

{ "login": "a-maci", "id": 23125439, "node_id": "MDQ6VXNlcjIzMTI1NDM5", "avatar_url": "https://avatars.githubusercontent.com/u/23125439?v=4", "gravatar_id": "", "url": "https://api.github.com/users/a-maci", "html_url": "https://github.com/a-maci", "followers_url": "https://api.github.com/users/a-maci/followers", "following_url": "https://api.github.com/users/a-maci/following{/other_user}", "gists_url": "https://api.github.com/users/a-maci/gists{/gist_id}", "starred_url": "https://api.github.com/users/a-maci/starred{/owner}{/repo}", "subscriptions_url": "https://api.github.com/users/a-maci/subscriptions", "organizations_url": "https://api.github.com/users/a-maci/orgs", "repos_url": "https://api.github.com/users/a-maci/repos", "events_url": "https://api.github.com/users/a-maci/events{/privacy}", "received_events_url": "https://api.github.com/users/a-maci/received_events", "type": "User", "site_admin": false }

[ { "id": 1108649070, "node_id": "MDU6TGFiZWwxMTA4NjQ5MDcw", "url": "https://api.github.com/repos/huggingface/transformers/labels/Need%20more%20information", "name": "Need more information", "color": "d876e3", "default": false, "description": "Further information is requested" }, { "id": 1314768611, "node_id": "MDU6TGFiZWwxMzE0NzY4NjEx", "url": "https://api.github.com/repos/huggingface/transformers/labels/wontfix", "name": "wontfix", "color": "ffffff", "default": true, "description": null } ]

[]

[ "Hi, what fine-tuning script and model are you referring to?", "I would like to know how to restart / continue runs as well.\r\nI would like to fine tune on half data first, checkpoint it. Then restart and continue on the other half of the data.\r\n\r\nLike the `main` function in this finetuning script:\r\nhttps://github.com/huggingface/pytorch-pretrained-BERT/blob/3fc63f126ddf883ba9659f13ec046c3639db7b7e/examples/lm_finetuning/simple_lm_finetuning.py", "@thomwolf Hi. I was experimenting with run_squad.py on colab. I was able to train and checkpoint the model after every 50 steps. However, for some reason, the notebook crashed and did not resume training. Is there a way to load that checkpoint and resume training from that point onwards? ", "I am fine-tuning using run_glue.py on bert. Have a checkpoint that I would like to continue from since my run crashed. Also, what happens to the tensorboard event file? For example, if my checkpoint is at iteration 250 (and my checkpoint crashed at 290), will the Tensorboard event file be appended correctly???", "This issue has been automatically marked as stale because it has not had recent activity. It will be closed if no further activity occurs. Thank you for your contributions.\n", "I think the solution is to change the model name to the checkpoint directory. When using the `run_glue.py` example script I changed the parameter from `--model_name_or_path bert-base-uncased` to `--model_name_or_path ../my-output-dir/checkpoint-1600`", "> I think the solution is to change the model name to the checkpoint directory. When using the `run_glue.py` example script I changed the parameter from `--model_name_or_path bert-base-uncased` to `--model_name_or_path ../my-output-dir/checkpoint-1600`\r\n\r\nHi, this works but may I know what did you do the OURPUT-DIR? Keeping the same one while \"overwriting\" or starting a new one? Thanks!", "> I think the solution is to change the model name to the checkpoint directory. When using the `run_glue.py` example script I changed the parameter from `--model_name_or_path bert-base-uncased` to `--model_name_or_path ../my-output-dir/checkpoint-1600`\r\n\r\nHi, I tried this. The following error message shows: \"We assumed '/cluster/home/xiazhi/finetune_results_republican/checkpoint-1500' was a path, a model identifier, or url to a directory containing vocabulary files named ['vocab.json', 'merges.txt'] but couldn't find such vocabulary files at this path or url.\" But only after all epochs are done will the vocal.json and merges.txt be generated.\r\n", "@anniezhi I have the same problem. This makes training very difficult; anyone have any ideas re: how to save the tokenizer whenever the checkpoints are saved?", "@anniezhi I figured it out - if loading from a checkpoint, use the additional argument --tokenizer_name and provide the name of your tokenizer. Here's my helper bash script for reference :\r\n\r\n```\r\n#!/bin/bash\r\nconda activate transformers\r\n\r\ncd \"${HOME}/Desktop\"\r\nrm -rf \"./${1}\"\r\n\r\nTRAIN_FILE=\"/media/b/F:/debiased_archive_200.h5\"\r\n\r\n#Matt login key\r\nwandb login MY_API_KEY\r\n\r\npython bao-ai/training_flows/run_language_custom_modeling.py \\\r\n --output_dir=\"./${1}\" \\\r\n --tokenizer_name=gpt2 \\\r\n --model_name_or_path=\"${2}\" \\\r\n --block_size \"${3}\" \\\r\n --per_device_train_batch_size \"${4}\" \\\r\n --do_train \\\r\n --train_data_file=$TRAIN_FILE\\\r\n```", "If you're using the latest release (v3.1.0), the tokenizer should be saved as well, so there's no need to use the `--tokenizer_name` anymore.\r\n\r\nFor any version <3.1.0, @apteryxlabs's solution is the way to go!", "> \r\n\r\nBrowse parameters, resume_from_checkpoint=./\r\n\r\nNow, the code runs from checkpoint\r\n\r\n", "> \r\n\r\nBrowse parameters, resume_from_checkpoint=./\r\n\r\nNow, the code runs from checkpoint\r\n\r\n" ]

{ "url": "https://api.github.com/repos/huggingface/transformers/issues/509/reactions", "total_count": 0, "+1": 0, "-1": 0, "laugh": 0, "hooray": 0, "confused": 0, "heart": 0, "rocket": 0, "eyes": 0 }

{ "login": "SivilTaram", "id": 10275209, "node_id": "MDQ6VXNlcjEwMjc1MjA5", "avatar_url": "https://avatars.githubusercontent.com/u/10275209?v=4", "gravatar_id": "", "url": "https://api.github.com/users/SivilTaram", "html_url": "https://github.com/SivilTaram", "followers_url": "https://api.github.com/users/SivilTaram/followers", "following_url": "https://api.github.com/users/SivilTaram/following{/other_user}", "gists_url": "https://api.github.com/users/SivilTaram/gists{/gist_id}", "starred_url": "https://api.github.com/users/SivilTaram/starred{/owner}{/repo}", "subscriptions_url": "https://api.github.com/users/SivilTaram/subscriptions", "organizations_url": "https://api.github.com/users/SivilTaram/orgs", "repos_url": "https://api.github.com/users/SivilTaram/repos", "events_url": "https://api.github.com/users/SivilTaram/events{/privacy}", "received_events_url": "https://api.github.com/users/SivilTaram/received_events", "type": "User", "site_admin": false }

[]

[]

[ "Thanks for the PR. This is fixed now." ]

{ "url": "https://api.github.com/repos/huggingface/transformers/issues/508/reactions", "total_count": 0, "+1": 0, "-1": 0, "laugh": 0, "hooray": 0, "confused": 0, "heart": 0, "rocket": 0, "eyes": 0 }

{ "url": "https://api.github.com/repos/huggingface/transformers/pulls/508", "html_url": "https://github.com/huggingface/transformers/pull/508", "diff_url": "https://github.com/huggingface/transformers/pull/508.diff", "patch_url": "https://github.com/huggingface/transformers/pull/508.patch", "merged_at": null }

{ "login": "rohuns", "id": 17604744, "node_id": "MDQ6VXNlcjE3NjA0NzQ0", "avatar_url": "https://avatars.githubusercontent.com/u/17604744?v=4", "gravatar_id": "", "url": "https://api.github.com/users/rohuns", "html_url": "https://github.com/rohuns", "followers_url": "https://api.github.com/users/rohuns/followers", "following_url": "https://api.github.com/users/rohuns/following{/other_user}", "gists_url": "https://api.github.com/users/rohuns/gists{/gist_id}", "starred_url": "https://api.github.com/users/rohuns/starred{/owner}{/repo}", "subscriptions_url": "https://api.github.com/users/rohuns/subscriptions", "organizations_url": "https://api.github.com/users/rohuns/orgs", "repos_url": "https://api.github.com/users/rohuns/repos", "events_url": "https://api.github.com/users/rohuns/events{/privacy}", "received_events_url": "https://api.github.com/users/rohuns/received_events", "type": "User", "site_admin": false }

[ { "id": 1260952223, "node_id": "MDU6TGFiZWwxMjYwOTUyMjIz", "url": "https://api.github.com/repos/huggingface/transformers/labels/Discussion", "name": "Discussion", "color": "22870e", "default": false, "description": "Discussion on a topic (keep it focused or open a new issue though)" } ]

[]

[ "Hi rohuns, I was wondering what padding value have you used for the lm_labels, since the -1 specified in the docs doesn't work for me on GPT2LMHead model. See #577. ", "> Hi rohuns, I was wondering what padding value have you used for the lm_labels, since the -1 specified in the docs doesn't work for me on GPT2LMHead model. See #577.\r\n\r\nI had just used -1, can take a look at your stack trace and respond on that chat", "Also to close this issue it appears others also achieved similar performance on the MC task, more details on the thread issue #468 ", "> > Hi rohuns, I was wondering what padding value have you used for the lm_labels, since the -1 specified in the docs doesn't work for me on GPT2LMHead model. See #577.\r\n> \r\n> I had just used -1, can take a look at your stack trace and respond on that chat\r\n\r\nYes, please do have a look. Here is a toy example with a hand-coded dataset to prove that the -1 throws an error. It looks like it's a library issue.\r\n[gpt2_simplified.py.zip](https://github.com/huggingface/pytorch-pretrained-BERT/files/3144185/gpt2_simplified.py.zip)\r\n\r\nRegards,\r\nAdrian\r\n" ]

{ "url": "https://api.github.com/repos/huggingface/transformers/issues/507/reactions", "total_count": 0, "+1": 0, "-1": 0, "laugh": 0, "hooray": 0, "confused": 0, "heart": 0, "rocket": 0, "eyes": 0 }

{ "login": "ailzhang", "id": 5248122, "node_id": "MDQ6VXNlcjUyNDgxMjI=", "avatar_url": "https://avatars.githubusercontent.com/u/5248122?v=4", "gravatar_id": "", "url": "https://api.github.com/users/ailzhang", "html_url": "https://github.com/ailzhang", "followers_url": "https://api.github.com/users/ailzhang/followers", "following_url": "https://api.github.com/users/ailzhang/following{/other_user}", "gists_url": "https://api.github.com/users/ailzhang/gists{/gist_id}", "starred_url": "https://api.github.com/users/ailzhang/starred{/owner}{/repo}", "subscriptions_url": "https://api.github.com/users/ailzhang/subscriptions", "organizations_url": "https://api.github.com/users/ailzhang/orgs", "repos_url": "https://api.github.com/users/ailzhang/repos", "events_url": "https://api.github.com/users/ailzhang/events{/privacy}", "received_events_url": "https://api.github.com/users/ailzhang/received_events", "type": "User", "site_admin": false }

[]

[]

[ "Hi @ailzhang,\r\nThis is great! I went through it and it looks good to me.\r\n\r\nI guess we should update the `from_pretrained` method of the other models as well (like [here](https://github.com/huggingface/pytorch-pretrained-BERT/blob/19666dcb3bee3e379f1458e295869957aac8590c/pytorch_pretrained_bert/modeling_openai.py#L420))\r\n\r\nDo you want to have a look at the other models (GPT, GPT-2 and Transformer-XL) and add them to the `hubconf.py` as well ?", "Hi @thomwolf, thanks for the quick reply! Yea we definitely would like to add GPT and Transformer-XL models in. \r\n\r\nI can definitely add them in this PR myself. Alternatively one thing could be super helpful to us would be someone from your team try out implementing a few models using `torch.hub` interfaces and let us know if you see any bugs/issues from a repo owner perspective :D. Let me know which way you prefer, thanks! \r\n\r\n Another question is about cache dir, pytorch has move to comply with XDG specification about caching dirs(https://github.com/pytorch/pytorch/issues/14693). Detailed logic can be found here https://pytorch.org/docs/master/hub.html#where-are-my-downloaded-models-saved ( I will fix the doc formatting soon :P ) Are you interested in moving to be in the same place? Happy to help on it as well. \r\n", "@thomwolf Any update on this? ;) Thanks!", "Hi @ailzhang, sorry for the delay, here are some answers to your questions:\r\n\r\n- `torch.hub`: I can give it a try but the present week is fully packed. I'll see if I can free some time next week. If you want to see it reach `master` faster, I'm also fine with you adding the other models.\r\n\r\nOne question I have here is that the pretrained models cannot really be used without the associated tokenizers. How is this supposed to work with `torch.hub`? Can you give me an example of usage (like the one in the readme for instance)?\r\n\r\n- update to `cache dir`: XDG specification seems nice indeed. If you want to give it a try it would be a lot cleaner than the present caching setting I guess.\r\n\r\nRelated note: we (Sebastian Ruder, Matthew Peters, Swabha Swayamdipta and I) are preparing a [tutorial on Transfer Learning in NLP to be held at NAACL](https://naacl2019.org/program/tutorials). We'll show various frameworks in action. I'll will see if we can include a `torch.hub` example.", "@thomwolf \r\nNote that there's a tokenizer in hub already. Typically we'd prefer hub only contains models, but in this case we also includes tokenizer as it's a required part. \r\nThere's an example in docstring of BertTokenizer. Is this good enough? \r\n```\r\n >>> sentence = 'Hello, World!'\r\n >>> tokenizer = torch.hub.load('ailzhang/pytorch-pretrained-BERT:hubconf', 'bertTokenizer', 'bert-base-cased', do_basic_tokenize=False, force_reload=False)\r\n >>> toks = tokenizer.tokenize(sentence)\r\n ['Hello', '##,', 'World', '##!']\r\n >>> ids = tokenizer.convert_tokens_to_ids(toks)\r\n [8667, 28136, 1291, 28125]\r\n```\r\nMaybe we can merge this PR first if it looks good? \r\n", "Oh indeed, I missed the tokenizer.\r\nOk let's go with this PR!" ]

{ "url": "https://api.github.com/repos/huggingface/transformers/issues/506/reactions", "total_count": 0, "+1": 0, "-1": 0, "laugh": 0, "hooray": 0, "confused": 0, "heart": 0, "rocket": 0, "eyes": 0 }

{ "url": "https://api.github.com/repos/huggingface/transformers/pulls/506", "html_url": "https://github.com/huggingface/transformers/pull/506", "diff_url": "https://github.com/huggingface/transformers/pull/506.diff", "patch_url": "https://github.com/huggingface/transformers/pull/506.patch", "merged_at": 1556132361000 }

{ "login": "shashwath94", "id": 7631779, "node_id": "MDQ6VXNlcjc2MzE3Nzk=", "avatar_url": "https://avatars.githubusercontent.com/u/7631779?v=4", "gravatar_id": "", "url": "https://api.github.com/users/shashwath94", "html_url": "https://github.com/shashwath94", "followers_url": "https://api.github.com/users/shashwath94/followers", "following_url": "https://api.github.com/users/shashwath94/following{/other_user}", "gists_url": "https://api.github.com/users/shashwath94/gists{/gist_id}", "starred_url": "https://api.github.com/users/shashwath94/starred{/owner}{/repo}", "subscriptions_url": "https://api.github.com/users/shashwath94/subscriptions", "organizations_url": "https://api.github.com/users/shashwath94/orgs", "repos_url": "https://api.github.com/users/shashwath94/repos", "events_url": "https://api.github.com/users/shashwath94/events{/privacy}", "received_events_url": "https://api.github.com/users/shashwath94/received_events", "type": "User", "site_admin": false }

[]

[]

[ "Here's an example of text generation, picks second most likely word at each step\r\n\r\n```\r\ntokenizer = TransfoXLTokenizer.from_pretrained('transfo-xl-wt103')\r\nmodel = TransfoXLLMHeadModel.from_pretrained('transfo-xl-wt103')\r\nline = \"Cars were invented in\"\r\nline_tokenized = tokenizer.tokenize(line)\r\nline_indexed = tokenizer.convert_tokens_to_ids(line_tokenized)\r\ntokens_tensor = torch.tensor([line_indexed])\r\ntokens_tensor = tokens_tensor.to(device)\r\n\r\nmax_predictions = 50\r\nmems = None\r\nfor i in range(max_predictions):\r\n predictions, mems = model(tokens_tensor, mems=mems)\r\n predicted_index = torch.topk(predictions[0, -1, :],5)[1][1].item()\r\n predicted_token = tokenizer.convert_ids_to_tokens([predicted_index])[0]\r\n print(predicted_token)\r\n predicted_index = torch.tensor([[predicted_index]]).to(device)\r\n tokens_tensor = torch.cat((tokens_tensor, predicted_index), dim=1)\r\n```\r\nShould produce\r\n\r\n```\r\nBritain\r\nand\r\nAmerica\r\n,\r\nbut\r\nthe\r\nfirst\r\ntwo\r\ncars\r\nhad\r\nto\r\nhave\r\nbeen\r\na\r\n\"\r\nTurbo\r\n```", "Yeah figured it out. Thanks nevertheless @yaroslavvb !", "@yaroslavvb I think, there is a bug in the code, you shared \r\n`predicted_index = torch.topk(predictions[0, -1, :],5)[1][1].item()`why is it not `predicted_index = torch.topk(predictions[0, -1, :],5)[1][0].item()` or probably its not a bug \r\n", "@yaroslavvb Why in the text generation with Transformer-XL there is a loop over the number of predictions requested, like max_predictions? \r\n\r\nGiven a fixed input like line = \"Cars were invented in\", which is 21 characters or 4 words (depending if trained for character output or word output), say, why one cannot generate say the next 21 characters or 4 words directly from the T-XL output all at once? Then generate another set of 21 characters or 4 words again in the next iteration? \r\n\r\nI thought one advantage of the T-XL vs the vanilla Transformer was this ability to predict a whole next sequence without having to loop by adding character by character or word by word at the input? \r\n\r\nIsn't the T-XL trained by computing the loss over the whole input and whole target (label) without looping?\r\nThus why would it be different during text generation? To provide a more accurate context along the prediction by adding the previous prediction one by one?", "@shashwath94 Could you please post your fix, so that we can learn by example? Thanks. ", "@gussmith you could do it this way, but empirically the results are very bad. The model loss is trained to maximize probability of \"next token prediction\". What looks like loss over a loss over whole sequence is actually a parallelization trick to compute many \"next token prediction\" losses in a single pass." ]

{ "url": "https://api.github.com/repos/huggingface/transformers/issues/505/reactions", "total_count": 0, "+1": 0, "-1": 0, "laugh": 0, "hooray": 0, "confused": 0, "heart": 0, "rocket": 0, "eyes": 0 }

{ "login": "ailzhang", "id": 5248122, "node_id": "MDQ6VXNlcjUyNDgxMjI=", "avatar_url": "https://avatars.githubusercontent.com/u/5248122?v=4", "gravatar_id": "", "url": "https://api.github.com/users/ailzhang", "html_url": "https://github.com/ailzhang", "followers_url": "https://api.github.com/users/ailzhang/followers", "following_url": "https://api.github.com/users/ailzhang/following{/other_user}", "gists_url": "https://api.github.com/users/ailzhang/gists{/gist_id}", "starred_url": "https://api.github.com/users/ailzhang/starred{/owner}{/repo}", "subscriptions_url": "https://api.github.com/users/ailzhang/subscriptions", "organizations_url": "https://api.github.com/users/ailzhang/orgs", "repos_url": "https://api.github.com/users/ailzhang/repos", "events_url": "https://api.github.com/users/ailzhang/events{/privacy}", "received_events_url": "https://api.github.com/users/ailzhang/received_events", "type": "User", "site_admin": false }

[]

[]

[ "actually this is related to my current work, I will send a fix along with my PR." ]

{ "url": "https://api.github.com/repos/huggingface/transformers/issues/504/reactions", "total_count": 0, "+1": 0, "-1": 0, "laugh": 0, "hooray": 0, "confused": 0, "heart": 0, "rocket": 0, "eyes": 0 }

{ "login": "SivilTaram", "id": 10275209, "node_id": "MDQ6VXNlcjEwMjc1MjA5", "avatar_url": "https://avatars.githubusercontent.com/u/10275209?v=4", "gravatar_id": "", "url": "https://api.github.com/users/SivilTaram", "html_url": "https://github.com/SivilTaram", "followers_url": "https://api.github.com/users/SivilTaram/followers", "following_url": "https://api.github.com/users/SivilTaram/following{/other_user}", "gists_url": "https://api.github.com/users/SivilTaram/gists{/gist_id}", "starred_url": "https://api.github.com/users/SivilTaram/starred{/owner}{/repo}", "subscriptions_url": "https://api.github.com/users/SivilTaram/subscriptions", "organizations_url": "https://api.github.com/users/SivilTaram/orgs", "repos_url": "https://api.github.com/users/SivilTaram/repos", "events_url": "https://api.github.com/users/SivilTaram/events{/privacy}", "received_events_url": "https://api.github.com/users/SivilTaram/received_events", "type": "User", "site_admin": false }

[ { "id": 1314768611, "node_id": "MDU6TGFiZWwxMzE0NzY4NjEx", "url": "https://api.github.com/repos/huggingface/transformers/labels/wontfix", "name": "wontfix", "color": "ffffff", "default": true, "description": null } ]

[]

[ "This issue has been automatically marked as stale because it has not had recent activity. It will be closed if no further activity occurs. Thank you for your contributions.\n" ]

{ "url": "https://api.github.com/repos/huggingface/transformers/issues/503/reactions", "total_count": 0, "+1": 0, "-1": 0, "laugh": 0, "hooray": 0, "confused": 0, "heart": 0, "rocket": 0, "eyes": 0 }

{ "url": "https://api.github.com/repos/huggingface/transformers/pulls/503", "html_url": "https://github.com/huggingface/transformers/pull/503", "diff_url": "https://github.com/huggingface/transformers/pull/503.diff", "patch_url": "https://github.com/huggingface/transformers/pull/503.patch", "merged_at": null }

{ "login": "serenaklm", "id": 34397223, "node_id": "MDQ6VXNlcjM0Mzk3MjIz", "avatar_url": "https://avatars.githubusercontent.com/u/34397223?v=4", "gravatar_id": "", "url": "https://api.github.com/users/serenaklm", "html_url": "https://github.com/serenaklm", "followers_url": "https://api.github.com/users/serenaklm/followers", "following_url": "https://api.github.com/users/serenaklm/following{/other_user}", "gists_url": "https://api.github.com/users/serenaklm/gists{/gist_id}", "starred_url": "https://api.github.com/users/serenaklm/starred{/owner}{/repo}", "subscriptions_url": "https://api.github.com/users/serenaklm/subscriptions", "organizations_url": "https://api.github.com/users/serenaklm/orgs", "repos_url": "https://api.github.com/users/serenaklm/repos", "events_url": "https://api.github.com/users/serenaklm/events{/privacy}", "received_events_url": "https://api.github.com/users/serenaklm/received_events", "type": "User", "site_admin": false }

[ { "id": 1260952223, "node_id": "MDU6TGFiZWwxMjYwOTUyMjIz", "url": "https://api.github.com/repos/huggingface/transformers/labels/Discussion", "name": "Discussion", "color": "22870e", "default": false, "description": "Discussion on a topic (keep it focused or open a new issue though)" }, { "id": 1314768611, "node_id": "MDU6TGFiZWwxMzE0NzY4NjEx", "url": "https://api.github.com/repos/huggingface/transformers/labels/wontfix", "name": "wontfix", "color": "ffffff", "default": true, "description": null } ]

[]

[ "Not really.\r\nYou should build a new sub-class of `BertPreTrainedModel` which is identical to `BertModel`but send back self-attention values in addition to the hidden states.\r\n", "I see. Thank you! ", "Hi, \r\n\r\nJust to add on. If this is what I would be doing, would it be advisable to fine-tune the weights for the pretrained model? \r\n\r\nRegards", "Probably.\r\nIt depends on what's your final use-case.", "This issue has been automatically marked as stale because it has not had recent activity. It will be closed if no further activity occurs. Thank you for your contributions.\n" ]

{ "url": "https://api.github.com/repos/huggingface/transformers/issues/502/reactions", "total_count": 0, "+1": 0, "-1": 0, "laugh": 0, "hooray": 0, "confused": 0, "heart": 0, "rocket": 0, "eyes": 0 }

{ "login": "wasiahmad", "id": 17520413, "node_id": "MDQ6VXNlcjE3NTIwNDEz", "avatar_url": "https://avatars.githubusercontent.com/u/17520413?v=4", "gravatar_id": "", "url": "https://api.github.com/users/wasiahmad", "html_url": "https://github.com/wasiahmad", "followers_url": "https://api.github.com/users/wasiahmad/followers", "following_url": "https://api.github.com/users/wasiahmad/following{/other_user}", "gists_url": "https://api.github.com/users/wasiahmad/gists{/gist_id}", "starred_url": "https://api.github.com/users/wasiahmad/starred{/owner}{/repo}", "subscriptions_url": "https://api.github.com/users/wasiahmad/subscriptions", "organizations_url": "https://api.github.com/users/wasiahmad/orgs", "repos_url": "https://api.github.com/users/wasiahmad/repos", "events_url": "https://api.github.com/users/wasiahmad/events{/privacy}", "received_events_url": "https://api.github.com/users/wasiahmad/received_events", "type": "User", "site_admin": false }

[ { "id": 1108649053, "node_id": "MDU6TGFiZWwxMTA4NjQ5MDUz", "url": "https://api.github.com/repos/huggingface/transformers/labels/Help%20wanted", "name": "Help wanted", "color": "008672", "default": false, "description": "Extra attention is needed, help appreciated" } ]

[]

[ "I noticed the following snippet in the code. (which I have edited to solve my problem)\r\n\r\n if args.do_train and (args.local_rank == -1 or torch.distributed.get_rank() == 0):\r\n # Save a trained model, configuration and tokenizer\r\n model_to_save = model.module if hasattr(model, 'module') else model # Only save the model it-self\r\n\r\n # If we save using the predefined names, we can load using `from_pretrained`\r\n output_model_file = os.path.join(args.output_dir, WEIGHTS_NAME)\r\n output_config_file = os.path.join(args.output_dir, CONFIG_NAME)\r\n\r\n torch.save(model_to_save.state_dict(), output_model_file)\r\n model_to_save.config.to_json_file(output_config_file)\r\n tokenizer.save_vocabulary(args.output_dir)\r\n\r\n # Load a trained model and vocabulary that you have fine-tuned\r\n model = BertForQuestionAnswering.from_pretrained(args.output_dir)\r\n tokenizer = BertTokenizer.from_pretrained(args.output_dir, do_lower_case=args.do_lower_case)\r\n else:\r\n model = BertForQuestionAnswering.from_pretrained(args.bert_model)\r\n\r\n\r\nSo, if we want to load the fine-tuned model only for prediction, need to load it from `args.output_dir`. But the current code loads from `args.bert_model` when we use `squad.py` only for prediction.", "@wasiahmad tokenizer is not needed at prediction time?\r\n\r\nThanks\r\nMahesh", "need help in understanding how to get the model trained with SQuAD + my dataset. Once trained, how to use it for actual prediction.\r\n\r\nmodel : BERT Question Answering\r\n", "@Swathygsb \r\nhttps://github.com/kamalkraj/BERT-SQuAD\r\ninference on bert-squad model", "> @Swathygsb\r\n> https://github.com/kamalkraj/BERT-SQuAD\r\n> inference on bert-squad model\r\n\r\nthx for your sharing, and there is inference on bert-squad model by tensorflow?\r\n3Q~" ]

{ "url": "https://api.github.com/repos/huggingface/transformers/issues/501/reactions", "total_count": 0, "+1": 0, "-1": 0, "laugh": 0, "hooray": 0, "confused": 0, "heart": 0, "rocket": 0, "eyes": 0 }

{ "login": "thomwolf", "id": 7353373, "node_id": "MDQ6VXNlcjczNTMzNzM=", "avatar_url": "https://avatars.githubusercontent.com/u/7353373?v=4", "gravatar_id": "", "url": "https://api.github.com/users/thomwolf", "html_url": "https://github.com/thomwolf", "followers_url": "https://api.github.com/users/thomwolf/followers", "following_url": "https://api.github.com/users/thomwolf/following{/other_user}", "gists_url": "https://api.github.com/users/thomwolf/gists{/gist_id}", "starred_url": "https://api.github.com/users/thomwolf/starred{/owner}{/repo}", "subscriptions_url": "https://api.github.com/users/thomwolf/subscriptions", "organizations_url": "https://api.github.com/users/thomwolf/orgs", "repos_url": "https://api.github.com/users/thomwolf/repos", "events_url": "https://api.github.com/users/thomwolf/events{/privacy}", "received_events_url": "https://api.github.com/users/thomwolf/received_events", "type": "User", "site_admin": false }

[]

[]

[]

{ "url": "https://api.github.com/repos/huggingface/transformers/issues/500/reactions", "total_count": 0, "+1": 0, "-1": 0, "laugh": 0, "hooray": 0, "confused": 0, "heart": 0, "rocket": 0, "eyes": 0 }

{ "url": "https://api.github.com/repos/huggingface/transformers/pulls/500", "html_url": "https://github.com/huggingface/transformers/pull/500", "diff_url": "https://github.com/huggingface/transformers/pull/500.diff", "patch_url": "https://github.com/huggingface/transformers/pull/500.patch", "merged_at": 1555507335000 }

{ "login": "directv00", "id": 37170991, "node_id": "MDQ6VXNlcjM3MTcwOTkx", "avatar_url": "https://avatars.githubusercontent.com/u/37170991?v=4", "gravatar_id": "", "url": "https://api.github.com/users/directv00", "html_url": "https://github.com/directv00", "followers_url": "https://api.github.com/users/directv00/followers", "following_url": "https://api.github.com/users/directv00/following{/other_user}", "gists_url": "https://api.github.com/users/directv00/gists{/gist_id}", "starred_url": "https://api.github.com/users/directv00/starred{/owner}{/repo}", "subscriptions_url": "https://api.github.com/users/directv00/subscriptions", "organizations_url": "https://api.github.com/users/directv00/orgs", "repos_url": "https://api.github.com/users/directv00/repos", "events_url": "https://api.github.com/users/directv00/events{/privacy}", "received_events_url": "https://api.github.com/users/directv00/received_events", "type": "User", "site_admin": false }

[ { "id": 1108649070, "node_id": "MDU6TGFiZWwxMTA4NjQ5MDcw", "url": "https://api.github.com/repos/huggingface/transformers/labels/Need%20more%20information", "name": "Need more information", "color": "d876e3", "default": false, "description": "Further information is requested" } ]

[]

[ "Did you install pytorch-pretrained-bert as indicated in the README?\r\n`pip install pytorch_pretrained_bert`\r\n\r\nYou don't have to convert the checkpoints yourself, there are already converted.\r\n\r\nTry reading the installation and usage sections of the README.", "Of cause I installed,\r\nMore precisely, error code is slightly changed.\r\n```bash\r\nTraceback (most recent call last):\r\n File \"run_squad.py\", line 37, in <module>\r\n from pytorch_pretrained_bert.file_utils import PYTORCH_PRETRAINED_BERT_CACHE, WEIGHTS_NAME, CONFIG_NAME\r\nImportError: cannot import name WEIGHTS_NAME\r\n```", "Hmm you are right, the examples are compatible with `master` only now that we have a new token serialization. I guess we'll have to do a new release (0.6.2) today so everybody is on the same page.\r\nLet me do that.", "Actually, we'll wait for the merge of #506.\r\n\r\nIn the meantime you can install from source and it should work.", "Oh it's done immediately when I installed from source. Thanks.", "Ok great.\r\n\r\nJust a side note on writing messages in github: you should add triple-quotes like this: \\``` before and after the command line, errors and code you are pasting. This way it's easier to read.\r\n\r\nEx:\r\n\\```\r\npip install -e .\r\n\\```\r\n\r\nwill display like:\r\n```\r\npip install -e .\r\n```", "Good point(triple quotes).\r\nI didn't know what to do, but now I have it all.\r\nThanks.", "> Actually, we'll wait for the merge of #506.\r\n> \r\n> In the meantime you can install from source and it should work.\r\n\r\nhow to \"install from source\"?", "@YanZhangADS \r\n\r\nYou can install from source with this command below\r\n```\r\ngit clone https://github.com/huggingface/pytorch-pretrained-BERT.git\r\ncd pytorch-pretrained-BERT\r\npython setup.py install\r\n```", "Same problem with \"ImportError: cannot import name WEIGHTS_NAME\". However, after building **0.6.1** from source, I get: \r\n```\r\nfrom pytorch_pretrained_bert.optimization import BertAdam, warmup_linear\r\n ImportError: cannot import name 'warmup_linear'\r\n```\r\nI don't need the warmup, so I removed the import, but letting you guys know that this is an import error as well. Thanks!", "Thanks for that @dumitrescustefan, we're working on it in #518.\r\nI'm closing this issue for now as we start to deviate from the original discussion.", "I just built from source. I'm still getting the same error as in original issue.", "The version 0.4.0 doesn't give this issue.\r\npip install pytorch_pretrained_bert==0.4.0" ]

{ "url": "https://api.github.com/repos/huggingface/transformers/issues/499/reactions", "total_count": 4, "+1": 4, "-1": 0, "laugh": 0, "hooray": 0, "confused": 0, "heart": 0, "rocket": 0, "eyes": 0 }

{ "login": "thomwolf", "id": 7353373, "node_id": "MDQ6VXNlcjczNTMzNzM=", "avatar_url": "https://avatars.githubusercontent.com/u/7353373?v=4", "gravatar_id": "", "url": "https://api.github.com/users/thomwolf", "html_url": "https://github.com/thomwolf", "followers_url": "https://api.github.com/users/thomwolf/followers", "following_url": "https://api.github.com/users/thomwolf/following{/other_user}", "gists_url": "https://api.github.com/users/thomwolf/gists{/gist_id}", "starred_url": "https://api.github.com/users/thomwolf/starred{/owner}{/repo}", "subscriptions_url": "https://api.github.com/users/thomwolf/subscriptions", "organizations_url": "https://api.github.com/users/thomwolf/orgs", "repos_url": "https://api.github.com/users/thomwolf/repos", "events_url": "https://api.github.com/users/thomwolf/events{/privacy}", "received_events_url": "https://api.github.com/users/thomwolf/received_events", "type": "User", "site_admin": false }

[]

[]

[]

{ "url": "https://api.github.com/repos/huggingface/transformers/issues/498/reactions", "total_count": 0, "+1": 0, "-1": 0, "laugh": 0, "hooray": 0, "confused": 0, "heart": 0, "rocket": 0, "eyes": 0 }

{ "url": "https://api.github.com/repos/huggingface/transformers/pulls/498", "html_url": "https://github.com/huggingface/transformers/pull/498", "diff_url": "https://github.com/huggingface/transformers/pull/498.diff", "patch_url": "https://github.com/huggingface/transformers/pull/498.patch", "merged_at": 1555492002000 }

{ "login": "yaroslavvb", "id": 23068, "node_id": "MDQ6VXNlcjIzMDY4", "avatar_url": "https://avatars.githubusercontent.com/u/23068?v=4", "gravatar_id": "", "url": "https://api.github.com/users/yaroslavvb", "html_url": "https://github.com/yaroslavvb", "followers_url": "https://api.github.com/users/yaroslavvb/followers", "following_url": "https://api.github.com/users/yaroslavvb/following{/other_user}", "gists_url": "https://api.github.com/users/yaroslavvb/gists{/gist_id}", "starred_url": "https://api.github.com/users/yaroslavvb/starred{/owner}{/repo}", "subscriptions_url": "https://api.github.com/users/yaroslavvb/subscriptions", "organizations_url": "https://api.github.com/users/yaroslavvb/orgs", "repos_url": "https://api.github.com/users/yaroslavvb/repos", "events_url": "https://api.github.com/users/yaroslavvb/events{/privacy}", "received_events_url": "https://api.github.com/users/yaroslavvb/received_events", "type": "User", "site_admin": false }

[]

[]

[ "Yes, this should be fixed by #498." ]

{ "url": "https://api.github.com/repos/huggingface/transformers/issues/497/reactions", "total_count": 0, "+1": 0, "-1": 0, "laugh": 0, "hooray": 0, "confused": 0, "heart": 0, "rocket": 0, "eyes": 0 }

{ "login": "8enmann", "id": 1021104, "node_id": "MDQ6VXNlcjEwMjExMDQ=", "avatar_url": "https://avatars.githubusercontent.com/u/1021104?v=4", "gravatar_id": "", "url": "https://api.github.com/users/8enmann", "html_url": "https://github.com/8enmann", "followers_url": "https://api.github.com/users/8enmann/followers", "following_url": "https://api.github.com/users/8enmann/following{/other_user}", "gists_url": "https://api.github.com/users/8enmann/gists{/gist_id}", "starred_url": "https://api.github.com/users/8enmann/starred{/owner}{/repo}", "subscriptions_url": "https://api.github.com/users/8enmann/subscriptions", "organizations_url": "https://api.github.com/users/8enmann/orgs", "repos_url": "https://api.github.com/users/8enmann/repos", "events_url": "https://api.github.com/users/8enmann/events{/privacy}", "received_events_url": "https://api.github.com/users/8enmann/received_events", "type": "User", "site_admin": false }

[]

[]

[ "Indeed, thanks @8enmann!" ]

{ "url": "https://api.github.com/repos/huggingface/transformers/issues/496/reactions", "total_count": 0, "+1": 0, "-1": 0, "laugh": 0, "hooray": 0, "confused": 0, "heart": 0, "rocket": 0, "eyes": 0 }

{ "url": "https://api.github.com/repos/huggingface/transformers/pulls/496", "html_url": "https://github.com/huggingface/transformers/pull/496", "diff_url": "https://github.com/huggingface/transformers/pull/496.diff", "patch_url": "https://github.com/huggingface/transformers/pull/496.patch", "merged_at": 1555492134000 }

{ "login": "SudoSharma", "id": 18308855, "node_id": "MDQ6VXNlcjE4MzA4ODU1", "avatar_url": "https://avatars.githubusercontent.com/u/18308855?v=4", "gravatar_id": "", "url": "https://api.github.com/users/SudoSharma", "html_url": "https://github.com/SudoSharma", "followers_url": "https://api.github.com/users/SudoSharma/followers", "following_url": "https://api.github.com/users/SudoSharma/following{/other_user}", "gists_url": "https://api.github.com/users/SudoSharma/gists{/gist_id}", "starred_url": "https://api.github.com/users/SudoSharma/starred{/owner}{/repo}", "subscriptions_url": "https://api.github.com/users/SudoSharma/subscriptions", "organizations_url": "https://api.github.com/users/SudoSharma/orgs", "repos_url": "https://api.github.com/users/SudoSharma/repos", "events_url": "https://api.github.com/users/SudoSharma/events{/privacy}", "received_events_url": "https://api.github.com/users/SudoSharma/received_events", "type": "User", "site_admin": false }

[]

[]

[ "Ok, great, thanks @SudoSharma!", "While the outputs are the same between 1e10 and 1e4, I shouldn't expect the outputs between fp32 and fp16 to be the same, should I? I get different outputs between the two when doing unconditional/conditional generation with top_k=40 but even with top_k=1. Usually they're the same for a while and then deviate. This is with Apex installed, so using FusedLayerNorm.\r\n\r\nIf I turn on Apex's AMP with `from apex import amp; amp.init()` then they still deviate but after a longer time (I think it makes the attention nn.Softmax use fp32). Have to remove the `model.half()` call when using AMP.\r\n\r\nPerhaps it's not realistic to have the outputs be the same when fp16 errors in the \"past\" tensors are compounding as the sequence gets longer? But it is surprising to see them differ for top_k=1 (deterministic) since only the largest logit affects the output there.\r\n\r\nP.S. For my site it's been enormously helpful to have this PyTorch implementation. @thomwolf Thank you!", "Hi @AdamDanielKing,\r\nCongratulation on your demo!\r\nAre you using the updated API for apex Amp? (https://nvidia.github.io/apex/amp.html)\r\nAlso, we should discuss this in a new issue? At first, I thought this was related to this PR but I understand it's not, right?", "@thomwolf You're probably right that a new issue is best. I've created one at #602.\r\n\r\nThanks for pointing out I was using the old Apex API. Switching to the new one unfortunately didn't fix the issue though." ]

{ "url": "https://api.github.com/repos/huggingface/transformers/issues/495/reactions", "total_count": 0, "+1": 0, "-1": 0, "laugh": 0, "hooray": 0, "confused": 0, "heart": 0, "rocket": 0, "eyes": 0 }

{ "url": "https://api.github.com/repos/huggingface/transformers/pulls/495", "html_url": "https://github.com/huggingface/transformers/pull/495", "diff_url": "https://github.com/huggingface/transformers/pull/495.diff", "patch_url": "https://github.com/huggingface/transformers/pull/495.patch", "merged_at": 1555492237000 }

{ "login": "SudoSharma", "id": 18308855, "node_id": "MDQ6VXNlcjE4MzA4ODU1", "avatar_url": "https://avatars.githubusercontent.com/u/18308855?v=4", "gravatar_id": "", "url": "https://api.github.com/users/SudoSharma", "html_url": "https://github.com/SudoSharma", "followers_url": "https://api.github.com/users/SudoSharma/followers", "following_url": "https://api.github.com/users/SudoSharma/following{/other_user}", "gists_url": "https://api.github.com/users/SudoSharma/gists{/gist_id}", "starred_url": "https://api.github.com/users/SudoSharma/starred{/owner}{/repo}", "subscriptions_url": "https://api.github.com/users/SudoSharma/subscriptions", "organizations_url": "https://api.github.com/users/SudoSharma/orgs", "repos_url": "https://api.github.com/users/SudoSharma/repos", "events_url": "https://api.github.com/users/SudoSharma/events{/privacy}", "received_events_url": "https://api.github.com/users/SudoSharma/received_events", "type": "User", "site_admin": false }

[]

[]

[ "Thanks!" ]

{ "url": "https://api.github.com/repos/huggingface/transformers/issues/494/reactions", "total_count": 0, "+1": 0, "-1": 0, "laugh": 0, "hooray": 0, "confused": 0, "heart": 0, "rocket": 0, "eyes": 0 }

{ "url": "https://api.github.com/repos/huggingface/transformers/pulls/494", "html_url": "https://github.com/huggingface/transformers/pull/494", "diff_url": "https://github.com/huggingface/transformers/pull/494.diff", "patch_url": "https://github.com/huggingface/transformers/pull/494.patch", "merged_at": 1555492296000 }

{ "login": "heslowen", "id": 22348625, "node_id": "MDQ6VXNlcjIyMzQ4NjI1", "avatar_url": "https://avatars.githubusercontent.com/u/22348625?v=4", "gravatar_id": "", "url": "https://api.github.com/users/heslowen", "html_url": "https://github.com/heslowen", "followers_url": "https://api.github.com/users/heslowen/followers", "following_url": "https://api.github.com/users/heslowen/following{/other_user}", "gists_url": "https://api.github.com/users/heslowen/gists{/gist_id}", "starred_url": "https://api.github.com/users/heslowen/starred{/owner}{/repo}", "subscriptions_url": "https://api.github.com/users/heslowen/subscriptions", "organizations_url": "https://api.github.com/users/heslowen/orgs", "repos_url": "https://api.github.com/users/heslowen/repos", "events_url": "https://api.github.com/users/heslowen/events{/privacy}", "received_events_url": "https://api.github.com/users/heslowen/received_events", "type": "User", "site_admin": false }

[ { "id": 1260952223, "node_id": "MDU6TGFiZWwxMjYwOTUyMjIz", "url": "https://api.github.com/repos/huggingface/transformers/labels/Discussion", "name": "Discussion", "color": "22870e", "default": false, "description": "Discussion on a topic (keep it focused or open a new issue though)" }, { "id": 1314768611, "node_id": "MDU6TGFiZWwxMzE0NzY4NjEx", "url": "https://api.github.com/repos/huggingface/transformers/labels/wontfix", "name": "wontfix", "color": "ffffff", "default": true, "description": null } ]

[]

[ "Without fine-tuning, BERT features are usually less useful than plain GloVe or wrd2vec indeed.\r\nThey start to be interesting when you fine-tune a classifier on top of BERT.\r\n\r\nSee the recent study by Matthew Peters, Sebastian Ruder, Noah A. Smith ([To Tune or Not to Tune? Adapting Pretrained Representations to Diverse Tasks](https://arxiv.org/abs/1903.05987)) for some practical tips on that.", "thank you so much~", "@heslowen could you please share the code for extracting features in order to use them for learning a classifier? Thanks.", "@joistick11 you can find a demo in extract_features.py", "Could you please help me?\r\nI was using bert-as-service (https://github.com/hanxiao/bert-as-service) and there is model method `encode`, which accepts list and returns list of the same size, each element containing sentence embedding. All the elements of the same size. \r\n1. When I use extract_features.py, it returns embedding for each recognized symbol in the sentence from the specified layers. I mean, instead of sentence embedding it returns symbols embeddings. How should I use it, for instance, to train an SVM? I am using `bert-base-multilingual-cased`\r\n2. Which layer output should I use? Is it with index `-1`?\r\n\r\nThanks you very much!", "@joistick11 you want to embed a sentence to a vector?\r\n`all_encoder_layers, pooled_output = model(input_ids, token_type_ids=None, attention_mask=input_mask)` pooled_output may help you.\r\nI have no idea about using these features to train an SVM although I know the theory about SVM.\r\nFor the second question, please refer to thomwolf's answer.\r\nI used the top 4 encoders_layers, but I did not get a better result than using Glove ", "@heslowen Hello, would you please help me? For a sequence like [cls I have a dog.sep], when I input this to Bert and get the last hidden layer of sequence out, let’s say the output is “vector”, is the vector[0] embedding of cls, vector[1] embedding of I, etc. vector[-1] embedding of sep?", "@heslowen How did you extract features after training a classifier on top of BERT? I've been trying to do the same, but I'm unable to do so. \r\nDo I first follow run_classifier.py, and then extract the features from tf.Estimator?", "@rvoak I use pytorch. I did it as the demo in extract_featrues.py. it is easy to do that, you just need to load a tokenizer, a bert model, then tokenize your sentences, and then run the model to get the encoded_layers", "@RomanShen yes you're right\r\n", "@heslowen Thanks for your reply!", "This issue has been automatically marked as stale because it has not had recent activity. It will be closed if no further activity occurs. Thank you for your contributions.\n", "@heslowen sorry about my english, now i doing embedding for sentence task, i tuned with my corpus with this library, and i received config.json, vocab.txt and model.bin file, but in bert's original doc, can extract feature when load from ckpt tensorflow checkpoint. according to your answer, i must write feature_extraction for torch, that's right ? please help me", "@hungph-dev-ict Do you mind opening a new issue with your problem? I'll try and help you out.", "@LysandreJik Thank you for your help, I will find solution for my problem, it's use last hidden layer in bert mechanism, but if you have a better solution, can you help me ?\r\nSo i have more concerns about with my corpus, with this library code, use tokenizer from pretrained BERT model, so I want use only BasicTokenizer. Can you help me ? ", "How long should the extract_features.py take to complete?\r\n\r\nwhen using 'bert-large-uncased' it takes seconds however it writes a blank file.\r\nwhen using 'bert-base-uncased' its been running for over 30 mins.\r\n\r\nany advice?\r\n\r\nthe code I used:\r\n\r\n!python extract_features.py \\\r\n --input_file data/src_train.txt \\\r\n --output_file data/output1.jsonl \\\r\n --bert_model bert-base-uncased \\\r\n --layers -1\r\n", "You can look at what the BertForSequenceClassification model [https://github.com/huggingface/transformers/blob/3ba5470eb85464df62f324bea88e20da234c423f/pytorch_pretrained_bert/modeling.py#L867 ](url) does in it’s forward 139.\r\nThe pooled_output obtained from self.bert would seem to be the features you are looking for." ]

{ "url": "https://api.github.com/repos/huggingface/transformers/issues/493/reactions", "total_count": 0, "+1": 0, "-1": 0, "laugh": 0, "hooray": 0, "confused": 0, "heart": 0, "rocket": 0, "eyes": 0 }

{ "login": "RayXu14", "id": 22774575, "node_id": "MDQ6VXNlcjIyNzc0NTc1", "avatar_url": "https://avatars.githubusercontent.com/u/22774575?v=4", "gravatar_id": "", "url": "https://api.github.com/users/RayXu14", "html_url": "https://github.com/RayXu14", "followers_url": "https://api.github.com/users/RayXu14/followers", "following_url": "https://api.github.com/users/RayXu14/following{/other_user}", "gists_url": "https://api.github.com/users/RayXu14/gists{/gist_id}", "starred_url": "https://api.github.com/users/RayXu14/starred{/owner}{/repo}", "subscriptions_url": "https://api.github.com/users/RayXu14/subscriptions", "organizations_url": "https://api.github.com/users/RayXu14/orgs", "repos_url": "https://api.github.com/users/RayXu14/repos", "events_url": "https://api.github.com/users/RayXu14/events{/privacy}", "received_events_url": "https://api.github.com/users/RayXu14/received_events", "type": "User", "site_admin": false }

[]

[]

[ "Yes. We are reproducing the behavior of the original optimizer, see [here](https://github.com/google-research/bert/blob/master/optimization.py#L65).", "thanks~", "but why?", "I have the same question, but did this prove to be better? Or is it just to speed up calculations?" ]

{ "url": "https://api.github.com/repos/huggingface/transformers/issues/492/reactions", "total_count": 0, "+1": 0, "-1": 0, "laugh": 0, "hooray": 0, "confused": 0, "heart": 0, "rocket": 0, "eyes": 0 }

{ "login": "yaroslavvb", "id": 23068, "node_id": "MDQ6VXNlcjIzMDY4", "avatar_url": "https://avatars.githubusercontent.com/u/23068?v=4", "gravatar_id": "", "url": "https://api.github.com/users/yaroslavvb", "html_url": "https://github.com/yaroslavvb", "followers_url": "https://api.github.com/users/yaroslavvb/followers", "following_url": "https://api.github.com/users/yaroslavvb/following{/other_user}", "gists_url": "https://api.github.com/users/yaroslavvb/gists{/gist_id}", "starred_url": "https://api.github.com/users/yaroslavvb/starred{/owner}{/repo}", "subscriptions_url": "https://api.github.com/users/yaroslavvb/subscriptions", "organizations_url": "https://api.github.com/users/yaroslavvb/orgs", "repos_url": "https://api.github.com/users/yaroslavvb/repos", "events_url": "https://api.github.com/users/yaroslavvb/events{/privacy}", "received_events_url": "https://api.github.com/users/yaroslavvb/received_events", "type": "User", "site_admin": false }

[ { "id": 1260952223, "node_id": "MDU6TGFiZWwxMjYwOTUyMjIz", "url": "https://api.github.com/repos/huggingface/transformers/labels/Discussion", "name": "Discussion", "color": "22870e", "default": false, "description": "Discussion on a topic (keep it focused or open a new issue though)" } ]

[]

[ "Accuracy goes to 31% if I use [stop-word filter](https://github.com/cybertronai/bflm/blob/51908bdd15477a0cedfbd010d489f8d355443b6a/eval_lambada_slow.py#L62), still seems lower than expected ([predictions](https://s3.amazonaws.com/yaroslavvb2/data/lambada_predictions_stopword_filter.txt))\r\n", "Hi, I doubt it's a problem with the model. Usually the culprit is too find in the pre-processing logic.\r\n\r\nYour dataset seems to be pre-processed but Radford, Wu et al. says they are using a version without preprocessing (end of section 3.3). GPT-2 is likely sensitive to tokenization issues and the like.\r\n\r\nIf you want to check the model it-self, you could try comparing with the predictions of the Tensorflow version on a few lambada completions?", "Applying [detokenization](https://github.com/cybertronai/bflm/blob/d58a6860451ee2afa3688aff13d104ad74001ebe/eval_lambada_slow.py#L77) raises accuracy to 33.11%\r\n\r\nI spot checked a few errors against TF implementation and they give the same errors, so it seems likely the difference is due to eval protocol, rather than the checkpoint", "IMHO \"without pre-processing\" means taking the original dataset without modification, which is what I also did here.\r\n\r\nHowever in the original dataset, everything is tokenized. IE \"haven't\" was turned into \"have n't\"\r\nEither way, undoing this tokenization only has a improvement of 2%, so there must be some deeper underlying difference in the way OpenAI did their evaluation.\r\n", "Indeed. It's not very clear to me what they mean exactly by \"stop-word filter\". It seems like the kind of heuristic that can have a very large impact on the performances.\r\n\r\nMaybe a better filtering is key. I would probably go with a sort of beam-search to compute the probability of having a punctuation/end-of-sentence token after the predicted word and use that to filter the results.", "I spoke with Alec and turns out for evaluation they got used the \"raw\" lambada corpus which was obtained by finding original sentences in book corpus that matched the tokenized versions in the lambada release. So to to reproduce the numbers we need the \"raw\" corpus https://github.com/openai/gpt-2/issues/131", "I'm now able to get within 1% of their reported accuracy on GPT2-small. The two missing modifications were:\r\n1. Evaluate on OpenAI's version of lambada which adds extra formatting\r\n2. Evaluate by counting number of times the last BPE token is predicted incorrectly instead of last word, details are in https://github.com/openai/gpt-2/issues/131#issuecomment-497136199" ]

{ "url": "https://api.github.com/repos/huggingface/transformers/issues/491/reactions", "total_count": 1, "+1": 1, "-1": 0, "laugh": 0, "hooray": 0, "confused": 0, "heart": 0, "rocket": 0, "eyes": 0 }

{ "login": "thomwolf", "id": 7353373, "node_id": "MDQ6VXNlcjczNTMzNzM=", "avatar_url": "https://avatars.githubusercontent.com/u/7353373?v=4", "gravatar_id": "", "url": "https://api.github.com/users/thomwolf", "html_url": "https://github.com/thomwolf", "followers_url": "https://api.github.com/users/thomwolf/followers", "following_url": "https://api.github.com/users/thomwolf/following{/other_user}", "gists_url": "https://api.github.com/users/thomwolf/gists{/gist_id}", "starred_url": "https://api.github.com/users/thomwolf/starred{/owner}{/repo}", "subscriptions_url": "https://api.github.com/users/thomwolf/subscriptions", "organizations_url": "https://api.github.com/users/thomwolf/orgs", "repos_url": "https://api.github.com/users/thomwolf/repos", "events_url": "https://api.github.com/users/thomwolf/events{/privacy}", "received_events_url": "https://api.github.com/users/thomwolf/received_events", "type": "User", "site_admin": false }

[]

[]

[]

{ "url": "https://api.github.com/repos/huggingface/transformers/issues/490/reactions", "total_count": 0, "+1": 0, "-1": 0, "laugh": 0, "hooray": 0, "confused": 0, "heart": 0, "rocket": 0, "eyes": 0 }

{ "url": "https://api.github.com/repos/huggingface/transformers/pulls/490", "html_url": "https://github.com/huggingface/transformers/pull/490", "diff_url": "https://github.com/huggingface/transformers/pull/490.diff", "patch_url": "https://github.com/huggingface/transformers/pull/490.patch", "merged_at": 1555337691000 }

{ "login": "thomwolf", "id": 7353373, "node_id": "MDQ6VXNlcjczNTMzNzM=", "avatar_url": "https://avatars.githubusercontent.com/u/7353373?v=4", "gravatar_id": "", "url": "https://api.github.com/users/thomwolf", "html_url": "https://github.com/thomwolf", "followers_url": "https://api.github.com/users/thomwolf/followers", "following_url": "https://api.github.com/users/thomwolf/following{/other_user}", "gists_url": "https://api.github.com/users/thomwolf/gists{/gist_id}", "starred_url": "https://api.github.com/users/thomwolf/starred{/owner}{/repo}", "subscriptions_url": "https://api.github.com/users/thomwolf/subscriptions", "organizations_url": "https://api.github.com/users/thomwolf/orgs", "repos_url": "https://api.github.com/users/thomwolf/repos", "events_url": "https://api.github.com/users/thomwolf/events{/privacy}", "received_events_url": "https://api.github.com/users/thomwolf/received_events", "type": "User", "site_admin": false }

[]

[]

[]

{ "url": "https://api.github.com/repos/huggingface/transformers/issues/489/reactions", "total_count": 0, "+1": 0, "-1": 0, "laugh": 0, "hooray": 0, "confused": 0, "heart": 0, "rocket": 0, "eyes": 0 }

{ "url": "https://api.github.com/repos/huggingface/transformers/pulls/489", "html_url": "https://github.com/huggingface/transformers/pull/489", "diff_url": "https://github.com/huggingface/transformers/pull/489.diff", "patch_url": "https://github.com/huggingface/transformers/pull/489.patch", "merged_at": 1555397395000 }

{ "login": "dhpollack", "id": 368699, "node_id": "MDQ6VXNlcjM2ODY5OQ==", "avatar_url": "https://avatars.githubusercontent.com/u/368699?v=4", "gravatar_id": "", "url": "https://api.github.com/users/dhpollack", "html_url": "https://github.com/dhpollack", "followers_url": "https://api.github.com/users/dhpollack/followers", "following_url": "https://api.github.com/users/dhpollack/following{/other_user}", "gists_url": "https://api.github.com/users/dhpollack/gists{/gist_id}", "starred_url": "https://api.github.com/users/dhpollack/starred{/owner}{/repo}", "subscriptions_url": "https://api.github.com/users/dhpollack/subscriptions", "organizations_url": "https://api.github.com/users/dhpollack/orgs", "repos_url": "https://api.github.com/users/dhpollack/repos", "events_url": "https://api.github.com/users/dhpollack/events{/privacy}", "received_events_url": "https://api.github.com/users/dhpollack/received_events", "type": "User", "site_admin": false }

[]

[]

[ "Indeed, it looks better.\r\nDo you want to have a look and confirm @rodgzilla?", "@thomwolf any word on this?", "Oh yes sorry. Looking at it and reading Alec Radford's paper on GPT (section 3.3) again, I think @rodgzilla was actually right in the original implementation.\r\n\r\nSo I guess we should close this PR.\r\n\r\nI still would have been happy to get @rodgzilla input on that.", "Oh sorry, we should still keep the `token_type_ids` and `attention_mask` `NoneType` fixes.\r\nThese ones are correct!" ]

{ "url": "https://api.github.com/repos/huggingface/transformers/issues/488/reactions", "total_count": 0, "+1": 0, "-1": 0, "laugh": 0, "hooray": 0, "confused": 0, "heart": 0, "rocket": 0, "eyes": 0 }

{ "url": "https://api.github.com/repos/huggingface/transformers/pulls/488", "html_url": "https://github.com/huggingface/transformers/pull/488", "diff_url": "https://github.com/huggingface/transformers/pull/488.diff", "patch_url": "https://github.com/huggingface/transformers/pull/488.patch", "merged_at": 1556219057000 }

{ "login": "ramild", "id": 9999944, "node_id": "MDQ6VXNlcjk5OTk5NDQ=", "avatar_url": "https://avatars.githubusercontent.com/u/9999944?v=4", "gravatar_id": "", "url": "https://api.github.com/users/ramild", "html_url": "https://github.com/ramild", "followers_url": "https://api.github.com/users/ramild/followers", "following_url": "https://api.github.com/users/ramild/following{/other_user}", "gists_url": "https://api.github.com/users/ramild/gists{/gist_id}", "starred_url": "https://api.github.com/users/ramild/starred{/owner}{/repo}", "subscriptions_url": "https://api.github.com/users/ramild/subscriptions", "organizations_url": "https://api.github.com/users/ramild/orgs", "repos_url": "https://api.github.com/users/ramild/repos", "events_url": "https://api.github.com/users/ramild/events{/privacy}", "received_events_url": "https://api.github.com/users/ramild/received_events", "type": "User", "site_admin": false }

[ { "id": 1108649053, "node_id": "MDU6TGFiZWwxMTA4NjQ5MDUz", "url": "https://api.github.com/repos/huggingface/transformers/labels/Help%20wanted", "name": "Help wanted", "color": "008672", "default": false, "description": "Extra attention is needed, help appreciated" }, { "id": 1260952223, "node_id": "MDU6TGFiZWwxMjYwOTUyMjIz", "url": "https://api.github.com/repos/huggingface/transformers/labels/Discussion", "name": "Discussion", "color": "22870e", "default": false, "description": "Discussion on a topic (keep it focused or open a new issue though)" }, { "id": 1314768611, "node_id": "MDU6TGFiZWwxMzE0NzY4NjEx", "url": "https://api.github.com/repos/huggingface/transformers/labels/wontfix", "name": "wontfix", "color": "ffffff", "default": true, "description": null } ]

[]

[ "UPD. I tried it with bert-multilingual-cased, but the results are still bad. A number of very simple (text, translated text) give very different probability distributions (the translated versions almost always fall into one major category).\r\n\r\nSpecifiically, **I fine-tune pre-trained bert-multilingual-cased on Russian text classification problem and then make a prediction using the model on an English text** (tried other languages -- nothing works).", "Hi, my feeling is that this is still an open research problem.\r\n\r\n[Here](https://twitter.com/nlpmattg/status/1091367511117881345) is a recent thread discussing the related problem of fine-tuning BERT on English SQuAD and trying to do QA in another language. Maybe you can get a pre-print from the RecitalAI guys if they haven't published it yet.", "This issue has been automatically marked as stale because it has not had recent activity. It will be closed if no further activity occurs. Thank you for your contributions.\n" ]

{ "url": "https://api.github.com/repos/huggingface/transformers/issues/487/reactions", "total_count": 0, "+1": 0, "-1": 0, "laugh": 0, "hooray": 0, "confused": 0, "heart": 0, "rocket": 0, "eyes": 0 }

{ "login": "tcqiuyu", "id": 6031166, "node_id": "MDQ6VXNlcjYwMzExNjY=", "avatar_url": "https://avatars.githubusercontent.com/u/6031166?v=4", "gravatar_id": "", "url": "https://api.github.com/users/tcqiuyu", "html_url": "https://github.com/tcqiuyu", "followers_url": "https://api.github.com/users/tcqiuyu/followers", "following_url": "https://api.github.com/users/tcqiuyu/following{/other_user}", "gists_url": "https://api.github.com/users/tcqiuyu/gists{/gist_id}", "starred_url": "https://api.github.com/users/tcqiuyu/starred{/owner}{/repo}", "subscriptions_url": "https://api.github.com/users/tcqiuyu/subscriptions", "organizations_url": "https://api.github.com/users/tcqiuyu/orgs", "repos_url": "https://api.github.com/users/tcqiuyu/repos", "events_url": "https://api.github.com/users/tcqiuyu/events{/privacy}", "received_events_url": "https://api.github.com/users/tcqiuyu/received_events", "type": "User", "site_admin": false }

[ { "id": 1260952223, "node_id": "MDU6TGFiZWwxMjYwOTUyMjIz", "url": "https://api.github.com/repos/huggingface/transformers/labels/Discussion", "name": "Discussion", "color": "22870e", "default": false, "description": "Discussion on a topic (keep it focused or open a new issue though)" }, { "id": 1314768611, "node_id": "MDU6TGFiZWwxMzE0NzY4NjEx", "url": "https://api.github.com/repos/huggingface/transformers/labels/wontfix", "name": "wontfix", "color": "ffffff", "default": true, "description": null } ]

[]

[ "Yes there are some optimizations. Please take a look here:\r\n\r\nhttps://hanxiao.github.io/2019/01/02/Serving-Google-BERT-in-Production-using-Tensorflow-and-ZeroMQ/#engineering-building-a-scalable-service", "Hi, there is no specific relation between the present repo (which provides PyTorch implementations of several transformer's models) and bert-as-a-service (which is built on top of BERT TensorFlow implementation AFAICT).\r\n\r\nWe want to keep the code simple in the present repo and don't provide specific optimizations out-of-the-box but if you want to add some of bert-as-a-service optimizations, here are the most straightforward ones:\r\n- use `with torch.no_grad()` to avoid computing gradient during evaluation (should divide memory consumption by two)\r\n- use fp16 code to get another factor of 2, probably the easiest is to use the amp wrapper in NVIDIA's apex library, see the details [here](https://github.com/NVIDIA/apex).", "This issue has been automatically marked as stale because it has not had recent activity. It will be closed if no further activity occurs. Thank you for your contributions.\n" ]

{ "url": "https://api.github.com/repos/huggingface/transformers/issues/486/reactions", "total_count": 0, "+1": 0, "-1": 0, "laugh": 0, "hooray": 0, "confused": 0, "heart": 0, "rocket": 0, "eyes": 0 }

{ "login": "KavyaGujjala", "id": 28920687, "node_id": "MDQ6VXNlcjI4OTIwNjg3", "avatar_url": "https://avatars.githubusercontent.com/u/28920687?v=4", "gravatar_id": "", "url": "https://api.github.com/users/KavyaGujjala", "html_url": "https://github.com/KavyaGujjala", "followers_url": "https://api.github.com/users/KavyaGujjala/followers", "following_url": "https://api.github.com/users/KavyaGujjala/following{/other_user}", "gists_url": "https://api.github.com/users/KavyaGujjala/gists{/gist_id}", "starred_url": "https://api.github.com/users/KavyaGujjala/starred{/owner}{/repo}", "subscriptions_url": "https://api.github.com/users/KavyaGujjala/subscriptions", "organizations_url": "https://api.github.com/users/KavyaGujjala/orgs", "repos_url": "https://api.github.com/users/KavyaGujjala/repos", "events_url": "https://api.github.com/users/KavyaGujjala/events{/privacy}", "received_events_url": "https://api.github.com/users/KavyaGujjala/received_events", "type": "User", "site_admin": false }

[]

[]

[ "I found out the issue. The text corpus I was using is just one document. So the code is for two or more documents only?", "Yes only for multiple documents.\r\nWe have a test now to check that since #478 thanks to @Rocketknight1." ]

{ "url": "https://api.github.com/repos/huggingface/transformers/issues/485/reactions", "total_count": 0, "+1": 0, "-1": 0, "laugh": 0, "hooray": 0, "confused": 0, "heart": 0, "rocket": 0, "eyes": 0 }

{ "login": "wasiahmad", "id": 17520413, "node_id": "MDQ6VXNlcjE3NTIwNDEz", "avatar_url": "https://avatars.githubusercontent.com/u/17520413?v=4", "gravatar_id": "", "url": "https://api.github.com/users/wasiahmad", "html_url": "https://github.com/wasiahmad", "followers_url": "https://api.github.com/users/wasiahmad/followers", "following_url": "https://api.github.com/users/wasiahmad/following{/other_user}", "gists_url": "https://api.github.com/users/wasiahmad/gists{/gist_id}", "starred_url": "https://api.github.com/users/wasiahmad/starred{/owner}{/repo}", "subscriptions_url": "https://api.github.com/users/wasiahmad/subscriptions", "organizations_url": "https://api.github.com/users/wasiahmad/orgs", "repos_url": "https://api.github.com/users/wasiahmad/repos", "events_url": "https://api.github.com/users/wasiahmad/events{/privacy}", "received_events_url": "https://api.github.com/users/wasiahmad/received_events", "type": "User", "site_admin": false }

[ { "id": 1108649070, "node_id": "MDU6TGFiZWwxMTA4NjQ5MDcw", "url": "https://api.github.com/repos/huggingface/transformers/labels/Need%20more%20information", "name": "Need more information", "color": "d876e3", "default": false, "description": "Further information is requested" } ]

[]

[ "Hi @wasiahmad,\r\nThis should actually already been taken care of by the WordPieceTokenizer ([here](https://github.com/huggingface/pytorch-pretrained-BERT/blob/19666dcb3bee3e379f1458e295869957aac8590c/pytorch_pretrained_bert/tokenization.py#L357)).\r\nDo you have a simple example to share so I can try to reproduce the behavior?", "Hi @thomwolf ,\r\n I alse found the same question when using BertTokenizer class. \r\nCode likes:\r\n```\r\nsent = ['hi', 'mary']\r\ntokenizer = BertTokenizer.from_pretrained('bert-base-cased')\r\nmodel_bert = BertModel.from_pretrained('bert-base-cased')\r\ntokenizer.convert_tokens_to_ids(sent)\r\n```", "Hi @thomwolf \r\nI didn't record that token for which I encountered that error but that token was not an English word, it was more like a weird symbol which was probably able to bypass WordPieceTokenizer. I was using BERT as a feature extractor for MSMARCO v2 QA dataset when I encountered the error.", "@congjianluo you should not tokenize the sentence your self but use Bert tokenizer.\r\nPlease follow the usage example in the [readme](https://github.com/huggingface/pytorch-pretrained-BERT#bert).\r\nIn your case:\r\n```python\r\nsent = \"hi mary\"\r\ntokenizer = BertTokenizer.from_pretrained('bert-base-cased')\r\ntokens = tokenizer.tokenize(sent)\r\ntokens_ids = tokenizer.convert_tokens_to_ids(tokens)\r\n```", "@wasiahmad ok then I'm closing this issue.\r\nFeel free to re-open it if you have a (reproductible) example of such issue." ]

{ "url": "https://api.github.com/repos/huggingface/transformers/issues/484/reactions", "total_count": 0, "+1": 0, "-1": 0, "laugh": 0, "hooray": 0, "confused": 0, "heart": 0, "rocket": 0, "eyes": 0 }

{ "login": "Akhila-Yerukola", "id": 4477323, "node_id": "MDQ6VXNlcjQ0NzczMjM=", "avatar_url": "https://avatars.githubusercontent.com/u/4477323?v=4", "gravatar_id": "", "url": "https://api.github.com/users/Akhila-Yerukola", "html_url": "https://github.com/Akhila-Yerukola", "followers_url": "https://api.github.com/users/Akhila-Yerukola/followers", "following_url": "https://api.github.com/users/Akhila-Yerukola/following{/other_user}", "gists_url": "https://api.github.com/users/Akhila-Yerukola/gists{/gist_id}", "starred_url": "https://api.github.com/users/Akhila-Yerukola/starred{/owner}{/repo}", "subscriptions_url": "https://api.github.com/users/Akhila-Yerukola/subscriptions", "organizations_url": "https://api.github.com/users/Akhila-Yerukola/orgs", "repos_url": "https://api.github.com/users/Akhila-Yerukola/repos", "events_url": "https://api.github.com/users/Akhila-Yerukola/events{/privacy}", "received_events_url": "https://api.github.com/users/Akhila-Yerukola/received_events", "type": "User", "site_admin": false }

[ { "id": 1260952223, "node_id": "MDU6TGFiZWwxMjYwOTUyMjIz", "url": "https://api.github.com/repos/huggingface/transformers/labels/Discussion", "name": "Discussion", "color": "22870e", "default": false, "description": "Discussion on a topic (keep it focused or open a new issue though)" }, { "id": 1314768611, "node_id": "MDU6TGFiZWwxMzE0NzY4NjEx", "url": "https://api.github.com/repos/huggingface/transformers/labels/wontfix", "name": "wontfix", "color": "ffffff", "default": true, "description": null } ]

[]

[ "Are you using context length 256 by chance?\r\n\r\nOn WikiText-2 I see\r\n\r\nI see 29.17378 using context length 1024 and 43.3393707 using context length 256\r\nThis is close to OpenAI report result of 29.41\r\n(although I don't get why it doesn't match it exactly)\r\n", "@yaroslavvb \r\nNo, I'm using a context length of 1024. \r\n\r\nAlthough, when I evaluate it on wikitext-2, my numbers dont match with yours. Would it be possible for you to share what you've done? I believe that you might have normalized the loss by the number of tokens after the gpt-2 tokenization, and not by the number of tokens originally in wikitext-2. ", "Here's our loss [computation](https://github.com/cybertronai/bflm/blob/b6ba6d97c9ccdf2b12e104fbdcd0bed25ada7b68/train_gpt2.py#L163)\r\n\r\nWe don't normalize the loss here, just take the average output of `model(batch, lm_labels=batch)` where model is `GPT2LMHeadModel.from_pretrained(args.model_name_or_path)`\r\n\r\nNote wikitext-103 and wikitext-2 have the same test set, so you should get `29.17378` perplexity on either one", "@Akhila-Yerukola when we tried normalizing by the original token count, our ppl number got much further from the paper.\r\n\r\nI ran `wikitext-2-raw/wiki.train.raw` through the encoder and got 2417793 tokens. Then I split the original on spaces and got 2088678. \r\n```py\r\nmath.exp(math.log(29.41) * 2088678 / 2417793) # 18.56\r\n```\r\nWe got 30.4 ppl unnormalized and 28 after [detokenizing](https://github.com/cybertronai/bflm/blob/master/detokenizer.py), so we're still quite far from 18.56.\r\nAnother possibility is that for evaluation, instead of chunking the dataset and shoving it through, they pass it in 1 token at a time the same way they do for sample generation. This would significantly reduce the loss because the model wouldn't \"forget\" what it was doing at the beginning of each sequence. I'll try this later today.", "As a quick test, I calculated loss based on only the last 10 tokens of the sequence without changing the chunking. I got test loss of 23.9 and train 24.9.", "@yaroslavvb @8enmann \r\nI still think the numbers don't match the paper.\r\nThe Table 3 in the paper which contains Zero-shot results on many datasets has a \"SOTA\" row which correspond to the reported perplexity numbers on the test sets of the data sets. (I verified this for wikitext-2 and wikitext-103)\r\nWhen I run [this](https://github.com/cybertronai/bflm/blob/b6ba6d97c9ccdf2b12e104fbdcd0bed25ada7b68/train_gpt2.py) evaluation script, I get ppl of 23.85 on wikitext-2 (not raw), 28.186 on wikitext-103 (not raw), 29.01 on wikitext-103-raw (and wikitext-2-raw). None of these match the reported 117M gpt-2 model (which is the model available) numbers from the paper (29.41 for wikitext-2, 37.5 for wikitext-103)\r\nFrom the looks of it wikitext-103 is way off.", "You're right, our results are very different from the paper. Perhaps they used the detokenized non-raw version and also for the labels passed -1 for <UNK> tokens, which causes them not to be evaluated for loss. This, combined with the 1 at a time trick I mentioned 2 days ago, could bring PPL down to ~18. However that still leaves a big puzzle as to why the loss they reported on wikitext-103 is so much worse than wikitext-2.", "Related issue, Lambada numbers also don't match the paper (31% instead of reported 46%) https://github.com/huggingface/pytorch-pretrained-BERT/issues/491", "This issue has been automatically marked as stale because it has not had recent activity. It will be closed if no further activity occurs. Thank you for your contributions.\n", "> @yaroslavvb @8enmann\r\n> When I run [this](https://github.com/cybertronai/bflm/blob/b6ba6d97c9ccdf2b12e104fbdcd0bed25ada7b68/train_gpt2.py) evaluation script, I get ppl of 23.85 on wikitext-2 (not raw), 28.186 on wikitext-103 (not raw), 29.01 on wikitext-103-raw (and wikitext-2-raw). None of these match the reported 117M gpt-2 model (which is the model available) numbers from the paper (29.41 for wikitext-2, 37.5 for wikitext-103)\r\n\r\nIs it possible to know why 0.7 and 0.3? --> 0.7 * exp_average_loss + 0.3 * loss.item()\r\n", "This issue has been automatically marked as stale because it has not had recent activity. It will be closed if no further activity occurs. Thank you for your contributions.\n", "This discussion is old, so this may not be applicable anymore. But I'd like to offer a data point if it's still the case. I tried gpt2 and gpt2-medium on OpenWebText (tokenized with HuggingFace's corresponding tokenizer settings), and I got the ppl about 24 and 18, respectively, whereas the openai version of them is 17 and 13, respectively. This is good enough to say that I probably didn't make any catastrophic mistake, but there still is some gap, which may or may not explain the performance gap on other datasets." ]

{ "url": "https://api.github.com/repos/huggingface/transformers/issues/483/reactions", "total_count": 0, "+1": 0, "-1": 0, "laugh": 0, "hooray": 0, "confused": 0, "heart": 0, "rocket": 0, "eyes": 0 }

{ "login": "sdeva14", "id": 18130526, "node_id": "MDQ6VXNlcjE4MTMwNTI2", "avatar_url": "https://avatars.githubusercontent.com/u/18130526?v=4", "gravatar_id": "", "url": "https://api.github.com/users/sdeva14", "html_url": "https://github.com/sdeva14", "followers_url": "https://api.github.com/users/sdeva14/followers", "following_url": "https://api.github.com/users/sdeva14/following{/other_user}", "gists_url": "https://api.github.com/users/sdeva14/gists{/gist_id}", "starred_url": "https://api.github.com/users/sdeva14/starred{/owner}{/repo}", "subscriptions_url": "https://api.github.com/users/sdeva14/subscriptions", "organizations_url": "https://api.github.com/users/sdeva14/orgs", "repos_url": "https://api.github.com/users/sdeva14/repos", "events_url": "https://api.github.com/users/sdeva14/events{/privacy}", "received_events_url": "https://api.github.com/users/sdeva14/received_events", "type": "User", "site_admin": false }

[ { "id": 1108649070, "node_id": "MDU6TGFiZWwxMTA4NjQ5MDcw", "url": "https://api.github.com/repos/huggingface/transformers/labels/Need%20more%20information", "name": "Need more information", "color": "d876e3", "default": false, "description": "Further information is requested" } ]

[]

[ "Hi Sungho,\r\nThis should already be taken care of by the BertTokenizer which defaults on the `unk` token when a sub-word is not in the vocabulary (see [here](https://github.com/huggingface/pytorch-pretrained-BERT/blob/master/pytorch_pretrained_bert/tokenization.py#L372)).\r\nWhat BERT model are you using?\r\nDo you have a simple example you can share so I can try to reproduce the behavior?", "Hi Tomas,\r\n\r\nIt is really shame to describe, I think it is just my typo problem again :(\r\nPlease ignore this issue.\r\n\r\nYou are absolutely right, there is an exception handling.\r\nJust I loaded a wrong model in the BertModel after tokenizing.\r\n\"bert-base-cased\" in the model part, Instead of \"bert-base-uncased\", which has a smaller vocab size than \"uncased\".\r\n\r\nI am so sorry to waste your time, and appreciate your favor again! :)\r\n\r\nBest\r\nSungho", "Ok make sense, good to know.\r\nSo there is no problem in the end?", "Yep, only my embarrassing here :)\r\n\r\nThough there is no performance change, but I believe it is not a problem of pretrained model at all, but \r\n just different application.\r\n\r\nThus, no problem for implementation at all, I appreciate it!", "Ok, let's close the issue then. Feel free to open a new one if you have other problems.", "May I know why cased vocabulary is smaller than uncased? " ]

{ "url": "https://api.github.com/repos/huggingface/transformers/issues/482/reactions", "total_count": 0, "+1": 0, "-1": 0, "laugh": 0, "hooray": 0, "confused": 0, "heart": 0, "rocket": 0, "eyes": 0 }

{ "login": "bladedsupernova", "id": 46139490, "node_id": "MDQ6VXNlcjQ2MTM5NDkw", "avatar_url": "https://avatars.githubusercontent.com/u/46139490?v=4", "gravatar_id": "", "url": "https://api.github.com/users/bladedsupernova", "html_url": "https://github.com/bladedsupernova", "followers_url": "https://api.github.com/users/bladedsupernova/followers", "following_url": "https://api.github.com/users/bladedsupernova/following{/other_user}", "gists_url": "https://api.github.com/users/bladedsupernova/gists{/gist_id}", "starred_url": "https://api.github.com/users/bladedsupernova/starred{/owner}{/repo}", "subscriptions_url": "https://api.github.com/users/bladedsupernova/subscriptions", "organizations_url": "https://api.github.com/users/bladedsupernova/orgs", "repos_url": "https://api.github.com/users/bladedsupernova/repos", "events_url": "https://api.github.com/users/bladedsupernova/events{/privacy}", "received_events_url": "https://api.github.com/users/bladedsupernova/received_events", "type": "User", "site_admin": false }

[ { "id": 1260952223, "node_id": "MDU6TGFiZWwxMjYwOTUyMjIz", "url": "https://api.github.com/repos/huggingface/transformers/labels/Discussion", "name": "Discussion", "color": "22870e", "default": false, "description": "Discussion on a topic (keep it focused or open a new issue though)" }, { "id": 1314768611, "node_id": "MDU6TGFiZWwxMzE0NzY4NjEx", "url": "https://api.github.com/repos/huggingface/transformers/labels/wontfix", "name": "wontfix", "color": "ffffff", "default": true, "description": null } ]

[]

[ "I'm not sure I understand your question.\r\n\r\nBut one thing BERT can do is mask-answering indeed (guessing a word in the middle of a sentence).\r\n\r\nBERT is quite bad at doing sequence prediction at the end of an input because it's not trained on partial sentences.", "When BERT fills-in a MASK, does it always use a answer it has seen or does it generalize and create never-seen Q-As? Ex. \"The *dog* in the hat, by Dr. Souse the Cat\"", "This issue has been automatically marked as stale because it has not had recent activity. It will be closed if no further activity occurs. Thank you for your contributions.\n" ]

{ "url": "https://api.github.com/repos/huggingface/transformers/issues/481/reactions", "total_count": 0, "+1": 0, "-1": 0, "laugh": 0, "hooray": 0, "confused": 0, "heart": 0, "rocket": 0, "eyes": 0 }

{ "login": "mboyanov", "id": 1610015, "node_id": "MDQ6VXNlcjE2MTAwMTU=", "avatar_url": "https://avatars.githubusercontent.com/u/1610015?v=4", "gravatar_id": "", "url": "https://api.github.com/users/mboyanov", "html_url": "https://github.com/mboyanov", "followers_url": "https://api.github.com/users/mboyanov/followers", "following_url": "https://api.github.com/users/mboyanov/following{/other_user}", "gists_url": "https://api.github.com/users/mboyanov/gists{/gist_id}", "starred_url": "https://api.github.com/users/mboyanov/starred{/owner}{/repo}", "subscriptions_url": "https://api.github.com/users/mboyanov/subscriptions", "organizations_url": "https://api.github.com/users/mboyanov/orgs", "repos_url": "https://api.github.com/users/mboyanov/repos", "events_url": "https://api.github.com/users/mboyanov/events{/privacy}", "received_events_url": "https://api.github.com/users/mboyanov/received_events", "type": "User", "site_admin": false }

[]

[]

[ "Ok, let's go for that @mboyanov!", "Great!" ]

{ "url": "https://api.github.com/repos/huggingface/transformers/issues/480/reactions", "total_count": 0, "+1": 0, "-1": 0, "laugh": 0, "hooray": 0, "confused": 0, "heart": 0, "rocket": 0, "eyes": 0 }

{ "url": "https://api.github.com/repos/huggingface/transformers/pulls/480", "html_url": "https://github.com/huggingface/transformers/pull/480", "diff_url": "https://github.com/huggingface/transformers/pull/480.diff", "patch_url": "https://github.com/huggingface/transformers/pull/480.patch", "merged_at": 1555318645000 }

{ "login": "asad1996172", "id": 19806866, "node_id": "MDQ6VXNlcjE5ODA2ODY2", "avatar_url": "https://avatars.githubusercontent.com/u/19806866?v=4", "gravatar_id": "", "url": "https://api.github.com/users/asad1996172", "html_url": "https://github.com/asad1996172", "followers_url": "https://api.github.com/users/asad1996172/followers", "following_url": "https://api.github.com/users/asad1996172/following{/other_user}", "gists_url": "https://api.github.com/users/asad1996172/gists{/gist_id}", "starred_url": "https://api.github.com/users/asad1996172/starred{/owner}{/repo}", "subscriptions_url": "https://api.github.com/users/asad1996172/subscriptions", "organizations_url": "https://api.github.com/users/asad1996172/orgs", "repos_url": "https://api.github.com/users/asad1996172/repos", "events_url": "https://api.github.com/users/asad1996172/events{/privacy}", "received_events_url": "https://api.github.com/users/asad1996172/received_events", "type": "User", "site_admin": false }

[ { "id": 1260952223, "node_id": "MDU6TGFiZWwxMjYwOTUyMjIz", "url": "https://api.github.com/repos/huggingface/transformers/labels/Discussion", "name": "Discussion", "color": "22870e", "default": false, "description": "Discussion on a topic (keep it focused or open a new issue though)" }, { "id": 1314768611, "node_id": "MDU6TGFiZWwxMzE0NzY4NjEx", "url": "https://api.github.com/repos/huggingface/transformers/labels/wontfix", "name": "wontfix", "color": "ffffff", "default": true, "description": null } ]

[]

[ "the allen institute tool may help you....it has probabilities of next 10 words for when adding a next word....i think the prorbablities are shown....may just be the 10 words but the git code is available i think so may be what you want.", "The output of `GPT2LMHeadModel` are logits so you can just apply a softmax (or log-softmax for log probabilities) on them to get probabilities for each token.\r\n\r\nThen if you want to get probabilities for words, you will need to multiply (or add if you used a log-softmax) the probabilities of the sub-words in each word.\r\n\r\nMaybe @sebastianGehrmann you have some additional insights (or plan to release the code of GLTR, it's a great demo!)?", "Thanks for pinging me @thomwolf. GLTR actually uses this amazing repo in the backend, so you could just use our code in conjunction. \r\n\r\nWe are still working on the docu, but you can find all our code here: https://github.com/HendrikStrobelt/detecting-fake-text ", "This issue has been automatically marked as stale because it has not had recent activity. It will be closed if no further activity occurs. Thank you for your contributions.\n" ]

{ "url": "https://api.github.com/repos/huggingface/transformers/issues/479/reactions", "total_count": 0, "+1": 0, "-1": 0, "laugh": 0, "hooray": 0, "confused": 0, "heart": 0, "rocket": 0, "eyes": 0 }

{ "login": "Rocketknight1", "id": 12866554, "node_id": "MDQ6VXNlcjEyODY2NTU0", "avatar_url": "https://avatars.githubusercontent.com/u/12866554?v=4", "gravatar_id": "", "url": "https://api.github.com/users/Rocketknight1", "html_url": "https://github.com/Rocketknight1", "followers_url": "https://api.github.com/users/Rocketknight1/followers", "following_url": "https://api.github.com/users/Rocketknight1/following{/other_user}", "gists_url": "https://api.github.com/users/Rocketknight1/gists{/gist_id}", "starred_url": "https://api.github.com/users/Rocketknight1/starred{/owner}{/repo}", "subscriptions_url": "https://api.github.com/users/Rocketknight1/subscriptions", "organizations_url": "https://api.github.com/users/Rocketknight1/orgs", "repos_url": "https://api.github.com/users/Rocketknight1/repos", "events_url": "https://api.github.com/users/Rocketknight1/events{/privacy}", "received_events_url": "https://api.github.com/users/Rocketknight1/received_events", "type": "User", "site_admin": false }

[]

[]

[ "Looks good to me, thanks @Rocketknight1 " ]

{ "url": "https://api.github.com/repos/huggingface/transformers/issues/478/reactions", "total_count": 0, "+1": 0, "-1": 0, "laugh": 0, "hooray": 0, "confused": 0, "heart": 0, "rocket": 0, "eyes": 0 }

{ "url": "https://api.github.com/repos/huggingface/transformers/pulls/478", "html_url": "https://github.com/huggingface/transformers/pull/478", "diff_url": "https://github.com/huggingface/transformers/pull/478.diff", "patch_url": "https://github.com/huggingface/transformers/pull/478.patch", "merged_at": 1555318558000 }

{ "login": "Shashi456", "id": 18056781, "node_id": "MDQ6VXNlcjE4MDU2Nzgx", "avatar_url": "https://avatars.githubusercontent.com/u/18056781?v=4", "gravatar_id": "", "url": "https://api.github.com/users/Shashi456", "html_url": "https://github.com/Shashi456", "followers_url": "https://api.github.com/users/Shashi456/followers", "following_url": "https://api.github.com/users/Shashi456/following{/other_user}", "gists_url": "https://api.github.com/users/Shashi456/gists{/gist_id}", "starred_url": "https://api.github.com/users/Shashi456/starred{/owner}{/repo}", "subscriptions_url": "https://api.github.com/users/Shashi456/subscriptions", "organizations_url": "https://api.github.com/users/Shashi456/orgs", "repos_url": "https://api.github.com/users/Shashi456/repos", "events_url": "https://api.github.com/users/Shashi456/events{/privacy}", "received_events_url": "https://api.github.com/users/Shashi456/received_events", "type": "User", "site_admin": false }

[ { "id": 1260952223, "node_id": "MDU6TGFiZWwxMjYwOTUyMjIz", "url": "https://api.github.com/repos/huggingface/transformers/labels/Discussion", "name": "Discussion", "color": "22870e", "default": false, "description": "Discussion on a topic (keep it focused or open a new issue though)" }, { "id": 1314768611, "node_id": "MDU6TGFiZWwxMzE0NzY4NjEx", "url": "https://api.github.com/repos/huggingface/transformers/labels/wontfix", "name": "wontfix", "color": "ffffff", "default": true, "description": null } ]

[]

[ "So p_M(S) is just the output of the model right?\r\n\r\nFor p_u(S), I think the easiest is probably to use the empirical probabilities.\r\n`TransfoXLTokenizer` has a counter to store words frequencies [here](https://github.com/huggingface/pytorch-pretrained-BERT/blob/master/pytorch_pretrained_bert/tokenization_transfo_xl.py#L98) which should be populated in the \"pretrained\" tokenizer so I would use and normalize this to get unconditional probabilities for each word and then compute SLOR.", "@thomwolf So the code would look something like : \r\n```sentence = \"Where is the dog going\"\r\ns_ids = convert_sentence_to_ids(sentence)\r\nconfig = TransfoXLConfig()\r\nmodel = TransfoXLModel(config)\r\nlast_hidden_state, new_mems = model(input_ids)\r\nfor i in sentence : \r\n uni_prob = freq(token)/no_of_tokens\r\n sent_uni_prob * = uni_prob \r\nSLOR = (1/len(sentence))* last_hidden_state - sent_uni_prob \r\n```\r\n\r\nor am i mistaken about the idea that the last_hidden_state is the sentence probability i'm looking for", "This issue has been automatically marked as stale because it has not had recent activity. It will be closed if no further activity occurs. Thank you for your contributions.\n", "In case someone found this, I found an implementation that seems legit here https://github.com/sgondala/GoogleConceptualCaptioning/commit/981e1b61ca5f84052f6237402319714d7fe70b80." ]

{ "url": "https://api.github.com/repos/huggingface/transformers/issues/477/reactions", "total_count": 0, "+1": 0, "-1": 0, "laugh": 0, "hooray": 0, "confused": 0, "heart": 0, "rocket": 0, "eyes": 0 }

{ "login": "jiahuigeng", "id": 22496073, "node_id": "MDQ6VXNlcjIyNDk2MDcz", "avatar_url": "https://avatars.githubusercontent.com/u/22496073?v=4", "gravatar_id": "", "url": "https://api.github.com/users/jiahuigeng", "html_url": "https://github.com/jiahuigeng", "followers_url": "https://api.github.com/users/jiahuigeng/followers", "following_url": "https://api.github.com/users/jiahuigeng/following{/other_user}", "gists_url": "https://api.github.com/users/jiahuigeng/gists{/gist_id}", "starred_url": "https://api.github.com/users/jiahuigeng/starred{/owner}{/repo}", "subscriptions_url": "https://api.github.com/users/jiahuigeng/subscriptions", "organizations_url": "https://api.github.com/users/jiahuigeng/orgs", "repos_url": "https://api.github.com/users/jiahuigeng/repos", "events_url": "https://api.github.com/users/jiahuigeng/events{/privacy}", "received_events_url": "https://api.github.com/users/jiahuigeng/received_events", "type": "User", "site_admin": false }

[ { "id": 1108649070, "node_id": "MDU6TGFiZWwxMTA4NjQ5MDcw", "url": "https://api.github.com/repos/huggingface/transformers/labels/Need%20more%20information", "name": "Need more information", "color": "d876e3", "default": false, "description": "Further information is requested" }, { "id": 1314768611, "node_id": "MDU6TGFiZWwxMzE0NzY4NjEx", "url": "https://api.github.com/repos/huggingface/transformers/labels/wontfix", "name": "wontfix", "color": "ffffff", "default": true, "description": null } ]

[]

[ "What hardware are you using?\r\nWhat is your software configuration (versions of python, pytorch, pytorh-pretrained-bert, etc...)?", "I'm not so sure that this is the same problem that I experienced.\r\n\r\nIf you installed apex, the problem may be related to BertLayerNorm, because BertLayerNorm will use FusedLayerNorm of apex library. Thus, if you disable to use FusedLayerNorm of apex library, then the segmentation fault will disappear.\r\n\r\nThe other solution is that you should reinstall apex. Please refer to https://github.com/NVIDIA/apex/issues/156#issuecomment-465301976.", "This issue has been automatically marked as stale because it has not had recent activity. It will be closed if no further activity occurs. Thank you for your contributions.\n", "@cartopy reinstall apex works well thx", "This issue has been automatically marked as stale because it has not had recent activity. It will be closed if no further activity occurs. Thank you for your contributions.\n" ]

{ "url": "https://api.github.com/repos/huggingface/transformers/issues/476/reactions", "total_count": 0, "+1": 0, "-1": 0, "laugh": 0, "hooray": 0, "confused": 0, "heart": 0, "rocket": 0, "eyes": 0 }

{ "login": "Jacob-Ma", "id": 46578842, "node_id": "MDQ6VXNlcjQ2NTc4ODQy", "avatar_url": "https://avatars.githubusercontent.com/u/46578842?v=4", "gravatar_id": "", "url": "https://api.github.com/users/Jacob-Ma", "html_url": "https://github.com/Jacob-Ma", "followers_url": "https://api.github.com/users/Jacob-Ma/followers", "following_url": "https://api.github.com/users/Jacob-Ma/following{/other_user}", "gists_url": "https://api.github.com/users/Jacob-Ma/gists{/gist_id}", "starred_url": "https://api.github.com/users/Jacob-Ma/starred{/owner}{/repo}", "subscriptions_url": "https://api.github.com/users/Jacob-Ma/subscriptions", "organizations_url": "https://api.github.com/users/Jacob-Ma/orgs", "repos_url": "https://api.github.com/users/Jacob-Ma/repos", "events_url": "https://api.github.com/users/Jacob-Ma/events{/privacy}", "received_events_url": "https://api.github.com/users/Jacob-Ma/received_events", "type": "User", "site_admin": false }

[ { "id": 1108649053, "node_id": "MDU6TGFiZWwxMTA4NjQ5MDUz", "url": "https://api.github.com/repos/huggingface/transformers/labels/Help%20wanted", "name": "Help wanted", "color": "008672", "default": false, "description": "Extra attention is needed, help appreciated" }, { "id": 1108649070, "node_id": "MDU6TGFiZWwxMTA4NjQ5MDcw", "url": "https://api.github.com/repos/huggingface/transformers/labels/Need%20more%20information", "name": "Need more information", "color": "d876e3", "default": false, "description": "Further information is requested" } ]

[]

[ "There is some non-determinism in cuDNN. Try setting `torch.backends.cudnn.deterministic = True` in your code: with that plus the RNG seeding, you should be able to get deterministic results.", "Yes go with @Rocketknight1 suggestion.\r\nAlso check that you set model in eval mode to disable the DropOut modules before evaluating.", "Hi, thank you for you guys reply. \r\nI simplify the model to a simple MNIST problem to check if there are similar phenomena.\r\nI also posted in pytorch/examples github, but no one replies. I found that it seems that it is PyTorch's problem. \r\n\r\nCould you please take a look at these posts?\r\n\r\n[mnist](https://github.com/pytorch/examples/issues/542), in these post, Yes, you are right `torch.backends.cudnn.deterministic = True` could bring some consistency to the result. \r\n\r\nHowever, even though you set `torch.backends.cudnn.deterministic = True`, STILL, when you compare evaluate at each epoch and only make final evaluation gives different results, you could see in this post:\r\n[minis final evaluation VS evaluation each epoch](https://github.com/pytorch/examples/issues/543)\r\n\r\nWhat would you think? Is it indeed pyTorch's problem? I have code attached in the post for your convenience to run. Thank you very much.\r\n\r\n", "> There is some non-determinism in cuDNN. Try setting `torch.backends.cudnn.deterministic = True` in your code: with that plus the RNG seeding, you should be able to get deterministic results.\r\n\r\nHi, thank you for your reply, when I post this issue, I have searched a lot. In the code I have already set the random seeding like this:\r\n```\r\nuse_cuda = not args.no_cuda and torch.cuda.is_available()\r\n\r\n# set seed\r\nrandom.seed(args.seed)\r\nnp.random.seed(args.seed)\r\ntorch.manual_seed(args.seed)\r\nif use_cuda:\r\n torch.cuda.manual_seed_all(args.seed) # if got GPU also set this seed\r\n```\r\nand also put `torch.backends.cudnn.deterministic = True` at the front of the script. \r\n\r\nBut the behavior is still the same.\r\n\r\nWhat do you think of this mystic difference?\r\nMy this post shows the detail and have code attached to have a try:\r\n[minis final evaluation VS evaluation each epoch](https://github.com/pytorch/examples/issues/543)\r\n", "> Yes go with @Rocketknight1 suggestion.\r\n> Also check that you set model in eval mode to disable the DropOut modules before evaluating.\r\n\r\nThank you for your reply. \r\nI have checked that, in my original code. \r\nIn train() and test() fuction, I already set model.train() and model.eval() respectively. \r\n\r\nBut the behavior is still the same. \r\nWhat do you think of this mystic difference?\r\n\r\nMy this post shows the detail and have code attached to have a try:\r\n[minis final evaluation VS evaluation each epoch](https://github.com/pytorch/examples/issues/543)", "This is really a PyTorch/CUDA issue rather than an issue with this repo, but there is more information here: https://pytorch.org/docs/stable/notes/randomness.html\r\n\r\nYou can also try setting `torch.backends.cudnn.benchmark = False`", "@Rocketknight1 \r\nThank you for your reply. \r\nI also tried to add `torch.backends.cudnn.benchmark=False`, the behavior is still the same as above.\r\nYeah, it seems to be a PyTorch problem. \r\n", "The #question channel on the PyTorch slack is a good place to get quick answers on these stuff.\r\nPyTorch's forum is also great." ]

{ "url": "https://api.github.com/repos/huggingface/transformers/issues/475/reactions", "total_count": 0, "+1": 0, "-1": 0, "laugh": 0, "hooray": 0, "confused": 0, "heart": 0, "rocket": 0, "eyes": 0 }

{ "login": "jiesutd", "id": 9111828, "node_id": "MDQ6VXNlcjkxMTE4Mjg=", "avatar_url": "https://avatars.githubusercontent.com/u/9111828?v=4", "gravatar_id": "", "url": "https://api.github.com/users/jiesutd", "html_url": "https://github.com/jiesutd", "followers_url": "https://api.github.com/users/jiesutd/followers", "following_url": "https://api.github.com/users/jiesutd/following{/other_user}", "gists_url": "https://api.github.com/users/jiesutd/gists{/gist_id}", "starred_url": "https://api.github.com/users/jiesutd/starred{/owner}{/repo}", "subscriptions_url": "https://api.github.com/users/jiesutd/subscriptions", "organizations_url": "https://api.github.com/users/jiesutd/orgs", "repos_url": "https://api.github.com/users/jiesutd/repos", "events_url": "https://api.github.com/users/jiesutd/events{/privacy}", "received_events_url": "https://api.github.com/users/jiesutd/received_events", "type": "User", "site_admin": false }

[]

[]

[ "Ok, thanks @jiesutd!" ]

{ "url": "https://api.github.com/repos/huggingface/transformers/issues/474/reactions", "total_count": 0, "+1": 0, "-1": 0, "laugh": 0, "hooray": 0, "confused": 0, "heart": 0, "rocket": 0, "eyes": 0 }

{ "url": "https://api.github.com/repos/huggingface/transformers/pulls/474", "html_url": "https://github.com/huggingface/transformers/pull/474", "diff_url": "https://github.com/huggingface/transformers/pull/474.diff", "patch_url": "https://github.com/huggingface/transformers/pull/474.patch", "merged_at": 1555318609000 }

{ "login": "mdasadul", "id": 8009589, "node_id": "MDQ6VXNlcjgwMDk1ODk=", "avatar_url": "https://avatars.githubusercontent.com/u/8009589?v=4", "gravatar_id": "", "url": "https://api.github.com/users/mdasadul", "html_url": "https://github.com/mdasadul", "followers_url": "https://api.github.com/users/mdasadul/followers", "following_url": "https://api.github.com/users/mdasadul/following{/other_user}", "gists_url": "https://api.github.com/users/mdasadul/gists{/gist_id}", "starred_url": "https://api.github.com/users/mdasadul/starred{/owner}{/repo}", "subscriptions_url": "https://api.github.com/users/mdasadul/subscriptions", "organizations_url": "https://api.github.com/users/mdasadul/orgs", "repos_url": "https://api.github.com/users/mdasadul/repos", "events_url": "https://api.github.com/users/mdasadul/events{/privacy}", "received_events_url": "https://api.github.com/users/mdasadul/received_events", "type": "User", "site_admin": false }

[ { "id": 1314768611, "node_id": "MDU6TGFiZWwxMzE0NzY4NjEx", "url": "https://api.github.com/repos/huggingface/transformers/labels/wontfix", "name": "wontfix", "color": "ffffff", "default": true, "description": null } ]

[]

[ "Looks good to me. It's perplexity so lower is better.\r\nYou can do a `math.exp(loss.item())` and call you model in a `with torch.no_grad()` context to be a little cleaner.", "Oh no wait, you need to compare to the shifted inputs:\r\n`loss=model(tensor_input[:-1], lm_labels=tensor_input[1:])`\r\nIt's a causal model, it predicts the next token given the previous ones.", "I can see inside the ```class OpenAIGPTLMHeadModel(OpenAIGPTPreTrainedModel)``` this shifting is happening \r\n``` if lm_labels is not None:\r\n # Shift so that tokens < n predict n\r\n shift_logits = lm_logits[:, :-1].contiguous()\r\n shift_labels = lm_labels[:, 1:].contiguous()\r\n```\r\n\r\nDo I still need to use \r\n ```loss=model(tensor_input[:-1], lm_labels=tensor_input[1:])```", "Oh you are right, this has been added now with #404.\r\nSo the way you are doing looks fine to me.\r\nShifting the logics inside the model can a bit dangerous for the people who are used to train a causal model the usual way, I'll add a mention in the README.", "Thanks for your quick response. I can see there is a minor bug when I am trying to predict with a sentence which has one word. You can re create the error by using my above code.\r\n```\r\nsentence='Learn'\r\nscore(sentence)\r\n```", "Unfortunately, given the way the model is trained (without using a token indicating the beginning of a sentence), I would say it does not make sense to try to get a score for a sentence with only one word.", "How can we use this to get the probability of a particular token? So, for instance, let's say we have the following sentence. \r\n\"He was going home\"\r\nand we want to get the probability of \"home\" given the context \"he was going\"\r\nlike in GLTR tool by harvard nlp @thomwolf ", "@thomwolf If the shifting of the lm_labels matrix isn't necessary (before passing into the model's forward method) because of the internal logit shifting, should the preprocess code for finetuning GPT1 in RocStories be changed? At [https://github.com/huggingface/pytorch-pretrained-BERT/blob/master/examples/run_openai_gpt.py#L86](url), I believe the continuations are shifted over in lm_labels one relative to input_ids.", "Oh yes, of course! Do you want to submit a PR on that? Otherwise I'll take of it later.", "For sure! Will look at it soon.\n\nOn Thu, Apr 25, 2019 at 11:33 PM Thomas Wolf <[email protected]>\nwrote:\n\n> Oh yes, of course! Do you want to submit a PR on that? Otherwise I'll take\n> of it later.\n>\n> —\n> You are receiving this because you commented.\n> Reply to this email directly, view it on GitHub\n> <https://github.com/huggingface/pytorch-pretrained-BERT/issues/473#issuecomment-486942939>,\n> or mute the thread\n> <https://github.com/notifications/unsubscribe-auth/AC6UQICJ3ROXNOJXROIKYN3PSKO4LANCNFSM4HFJZIVQ>\n> .\n>\n", "@thomwolf Hey how can I give my own checkpoint files to the model while loading\r\n", "This issue has been automatically marked as stale because it has not had recent activity. It will be closed if no further activity occurs. Thank you for your contributions.\n", "Is this score normalized on sentence lenght? And if not, what do I need to change to normalize it? \r\n", "I'm confused whether the right way to calculate the perplexity for GPT2 is what the OP has done or as per the documentation https://huggingface.co./transformers/perplexity.html? Or both are equivalent for some value of the stride?" ]

{ "url": "https://api.github.com/repos/huggingface/transformers/issues/473/reactions", "total_count": 30, "+1": 29, "-1": 0, "laugh": 0, "hooray": 0, "confused": 0, "heart": 1, "rocket": 0, "eyes": 0 }

{ "login": "hongkahjun", "id": 29894605, "node_id": "MDQ6VXNlcjI5ODk0NjA1", "avatar_url": "https://avatars.githubusercontent.com/u/29894605?v=4", "gravatar_id": "", "url": "https://api.github.com/users/hongkahjun", "html_url": "https://github.com/hongkahjun", "followers_url": "https://api.github.com/users/hongkahjun/followers", "following_url": "https://api.github.com/users/hongkahjun/following{/other_user}", "gists_url": "https://api.github.com/users/hongkahjun/gists{/gist_id}", "starred_url": "https://api.github.com/users/hongkahjun/starred{/owner}{/repo}", "subscriptions_url": "https://api.github.com/users/hongkahjun/subscriptions", "organizations_url": "https://api.github.com/users/hongkahjun/orgs", "repos_url": "https://api.github.com/users/hongkahjun/repos", "events_url": "https://api.github.com/users/hongkahjun/events{/privacy}", "received_events_url": "https://api.github.com/users/hongkahjun/received_events", "type": "User", "site_admin": false }

[]

[]

[ "managed to install it by running \r\n\r\n> sudo apt-get install python3 python-dev" ]

{ "url": "https://api.github.com/repos/huggingface/transformers/issues/472/reactions", "total_count": 0, "+1": 0, "-1": 0, "laugh": 0, "hooray": 0, "confused": 0, "heart": 0, "rocket": 0, "eyes": 0 }

{ "login": "Jonbean", "id": 10235447, "node_id": "MDQ6VXNlcjEwMjM1NDQ3", "avatar_url": "https://avatars.githubusercontent.com/u/10235447?v=4", "gravatar_id": "", "url": "https://api.github.com/users/Jonbean", "html_url": "https://github.com/Jonbean", "followers_url": "https://api.github.com/users/Jonbean/followers", "following_url": "https://api.github.com/users/Jonbean/following{/other_user}", "gists_url": "https://api.github.com/users/Jonbean/gists{/gist_id}", "starred_url": "https://api.github.com/users/Jonbean/starred{/owner}{/repo}", "subscriptions_url": "https://api.github.com/users/Jonbean/subscriptions", "organizations_url": "https://api.github.com/users/Jonbean/orgs", "repos_url": "https://api.github.com/users/Jonbean/repos", "events_url": "https://api.github.com/users/Jonbean/events{/privacy}", "received_events_url": "https://api.github.com/users/Jonbean/received_events", "type": "User", "site_admin": false }

[]

[]

[ "Good catch, I'll fix this." ]

{ "url": "https://api.github.com/repos/huggingface/transformers/issues/471/reactions", "total_count": 0, "+1": 0, "-1": 0, "laugh": 0, "hooray": 0, "confused": 0, "heart": 0, "rocket": 0, "eyes": 0 }

{ "login": "junchen14", "id": 19342556, "node_id": "MDQ6VXNlcjE5MzQyNTU2", "avatar_url": "https://avatars.githubusercontent.com/u/19342556?v=4", "gravatar_id": "", "url": "https://api.github.com/users/junchen14", "html_url": "https://github.com/junchen14", "followers_url": "https://api.github.com/users/junchen14/followers", "following_url": "https://api.github.com/users/junchen14/following{/other_user}", "gists_url": "https://api.github.com/users/junchen14/gists{/gist_id}", "starred_url": "https://api.github.com/users/junchen14/starred{/owner}{/repo}", "subscriptions_url": "https://api.github.com/users/junchen14/subscriptions", "organizations_url": "https://api.github.com/users/junchen14/orgs", "repos_url": "https://api.github.com/users/junchen14/repos", "events_url": "https://api.github.com/users/junchen14/events{/privacy}", "received_events_url": "https://api.github.com/users/junchen14/received_events", "type": "User", "site_admin": false }

[ { "id": 1108649053, "node_id": "MDU6TGFiZWwxMTA4NjQ5MDUz", "url": "https://api.github.com/repos/huggingface/transformers/labels/Help%20wanted", "name": "Help wanted", "color": "008672", "default": false, "description": "Extra attention is needed, help appreciated" }, { "id": 1108649070, "node_id": "MDU6TGFiZWwxMTA4NjQ5MDcw", "url": "https://api.github.com/repos/huggingface/transformers/labels/Need%20more%20information", "name": "Need more information", "color": "d876e3", "default": false, "description": "Further information is requested" }, { "id": 1260952223, "node_id": "MDU6TGFiZWwxMjYwOTUyMjIz", "url": "https://api.github.com/repos/huggingface/transformers/labels/Discussion", "name": "Discussion", "color": "22870e", "default": false, "description": "Discussion on a topic (keep it focused or open a new issue though)" }, { "id": 1314768611, "node_id": "MDU6TGFiZWwxMzE0NzY4NjEx", "url": "https://api.github.com/repos/huggingface/transformers/labels/wontfix", "name": "wontfix", "color": "ffffff", "default": true, "description": null } ]

[]

[ "This issue has been automatically marked as stale because it has not had recent activity. It will be closed if no further activity occurs. Thank you for your contributions.\n", "I'm having a similar issue where I run a BERT tutorial by Chris McCormick and receive 65-78% accuracy on training data and 0.0% for test data. Was anyone able to diagnose a result of 0.0%? One theory I have is a division by zero error for very small numbers that are rounded to zero..." ]

{ "url": "https://api.github.com/repos/huggingface/transformers/issues/470/reactions", "total_count": 0, "+1": 0, "-1": 0, "laugh": 0, "hooray": 0, "confused": 0, "heart": 0, "rocket": 0, "eyes": 0 }

{ "login": "floAlpha", "id": 21286485, "node_id": "MDQ6VXNlcjIxMjg2NDg1", "avatar_url": "https://avatars.githubusercontent.com/u/21286485?v=4", "gravatar_id": "", "url": "https://api.github.com/users/floAlpha", "html_url": "https://github.com/floAlpha", "followers_url": "https://api.github.com/users/floAlpha/followers", "following_url": "https://api.github.com/users/floAlpha/following{/other_user}", "gists_url": "https://api.github.com/users/floAlpha/gists{/gist_id}", "starred_url": "https://api.github.com/users/floAlpha/starred{/owner}{/repo}", "subscriptions_url": "https://api.github.com/users/floAlpha/subscriptions", "organizations_url": "https://api.github.com/users/floAlpha/orgs", "repos_url": "https://api.github.com/users/floAlpha/repos", "events_url": "https://api.github.com/users/floAlpha/events{/privacy}", "received_events_url": "https://api.github.com/users/floAlpha/received_events", "type": "User", "site_admin": false }

[ { "id": 1314768611, "node_id": "MDU6TGFiZWwxMzE0NzY4NjEx", "url": "https://api.github.com/repos/huggingface/transformers/labels/wontfix", "name": "wontfix", "color": "ffffff", "default": true, "description": null } ]

[]

[ "First, yes you can. You should be able to download the model(whether use shell command or python downloader) into a specified directory and modify the loading lines. That works for me. \r\nSecond, please be appreciated to these fantastic works huggingface team pulled off. It's not their fault that you don't have stable access, so help yourself.", "Humm wording apart, maybe we could relax the internet connection check indeed, I'll have a look.", "Sincerely thank you for your reply, you are the great contributors, I apologize again for my inappropriate expression, mainly because I can`t properly solve the problem till midnight. I just want to make my own suggestion for people encountered this like me, so I opened this issue. Actually I am very grateful for your work, and appreciate the open source community. Best regards!", "Ok, network connection check has been relaxed in the now merged #500.\r\nIt will be included in the next PyPI release (probably next week).\r\nIn the meantime you can install from `master`.", "So nice you are，I cannot continue my research without pytorch-pretrained-BERT，Thank you for your Incredible devotion.With my sincere apology.", "This issue has been automatically marked as stale because it has not had recent activity. It will be closed if no further activity occurs. Thank you for your contributions.\n" ]

{ "url": "https://api.github.com/repos/huggingface/transformers/issues/469/reactions", "total_count": 0, "+1": 0, "-1": 0, "laugh": 0, "hooray": 0, "confused": 0, "heart": 0, "rocket": 0, "eyes": 0 }

{ "login": "Jonbean", "id": 10235447, "node_id": "MDQ6VXNlcjEwMjM1NDQ3", "avatar_url": "https://avatars.githubusercontent.com/u/10235447?v=4", "gravatar_id": "", "url": "https://api.github.com/users/Jonbean", "html_url": "https://github.com/Jonbean", "followers_url": "https://api.github.com/users/Jonbean/followers", "following_url": "https://api.github.com/users/Jonbean/following{/other_user}", "gists_url": "https://api.github.com/users/Jonbean/gists{/gist_id}", "starred_url": "https://api.github.com/users/Jonbean/starred{/owner}{/repo}", "subscriptions_url": "https://api.github.com/users/Jonbean/subscriptions", "organizations_url": "https://api.github.com/users/Jonbean/orgs", "repos_url": "https://api.github.com/users/Jonbean/repos", "events_url": "https://api.github.com/users/Jonbean/events{/privacy}", "received_events_url": "https://api.github.com/users/Jonbean/received_events", "type": "User", "site_admin": false }

[ { "id": 1314768611, "node_id": "MDU6TGFiZWwxMzE0NzY4NjEx", "url": "https://api.github.com/repos/huggingface/transformers/labels/wontfix", "name": "wontfix", "color": "ffffff", "default": true, "description": null } ]

[]

[ "What kind of task are you fine-tuning on? I If it's something like ROCstories task, you need the extra tokens. I think people are doing BERT for up-stream tasks because birectional context gives better results than left-to-right", "Hi @yaroslavvb, I am mostly focusing on classification tasks(like ROCstories as you mentioned). I just want to confirm that the special tokens used by GPT and GPT2 should be treated the same(add to the end of the vocabulary and feed in along with the original text). \r\n\r\nWe would like to run both BERT and GPT2 which are the two SOTA models, since they shine in different ways. \r\n\r\nI will be happy to submit a pull request if this is a valid extension to the code base or if anyone is interested. ", "Indeed we should probably add additional embeddings for GPT-2 also.\r\nI'll give it a look, should be pretty easy to add.", "I just tried implementing the changes suggested in order to make gpt-2 amenable to being finetuned on the Cloze Story Task from ROCStories. However, my eval-accuracy seems to be topping out at 68%. Is this what others are getting?", "I also want to fine tune gpt-2 for qa and run it on squad. I am new to the field. Should I be following BertForQuestionAnswering and run BERT or SQuAD as a model to do the same for gpt-2? ", "Hi @rohuns, I met the same situation, the performance of the pre-trained GPT-2 with extra task head is poor on both ROC and sts. I don't truly know the reason. \r\nMy hypothesis is GPT-2 was not trained in the same \"multi-task\" fashion as the GPT1, therefore adding the special token will destroy the model or at least having hard time generating good representation for the sentence for the downstream tasks. \r\nI hope someone can get the fine-tuning work to disprove the above hypothesis. \r\n", "This issue has been automatically marked as stale because it has not had recent activity. It will be closed if no further activity occurs. Thank you for your contributions.\n" ]

{ "url": "https://api.github.com/repos/huggingface/transformers/issues/468/reactions", "total_count": 2, "+1": 2, "-1": 0, "laugh": 0, "hooray": 0, "confused": 0, "heart": 0, "rocket": 0, "eyes": 0 }

{ "login": "yaroslavvb", "id": 23068, "node_id": "MDQ6VXNlcjIzMDY4", "avatar_url": "https://avatars.githubusercontent.com/u/23068?v=4", "gravatar_id": "", "url": "https://api.github.com/users/yaroslavvb", "html_url": "https://github.com/yaroslavvb", "followers_url": "https://api.github.com/users/yaroslavvb/followers", "following_url": "https://api.github.com/users/yaroslavvb/following{/other_user}", "gists_url": "https://api.github.com/users/yaroslavvb/gists{/gist_id}", "starred_url": "https://api.github.com/users/yaroslavvb/starred{/owner}{/repo}", "subscriptions_url": "https://api.github.com/users/yaroslavvb/subscriptions", "organizations_url": "https://api.github.com/users/yaroslavvb/orgs", "repos_url": "https://api.github.com/users/yaroslavvb/repos", "events_url": "https://api.github.com/users/yaroslavvb/events{/privacy}", "received_events_url": "https://api.github.com/users/yaroslavvb/received_events", "type": "User", "site_admin": false }

[]

[]

[ "Thanks Yaroslav!" ]

{ "url": "https://api.github.com/repos/huggingface/transformers/issues/467/reactions", "total_count": 0, "+1": 0, "-1": 0, "laugh": 0, "hooray": 0, "confused": 0, "heart": 0, "rocket": 0, "eyes": 0 }

{ "url": "https://api.github.com/repos/huggingface/transformers/pulls/467", "html_url": "https://github.com/huggingface/transformers/pull/467", "diff_url": "https://github.com/huggingface/transformers/pull/467.diff", "patch_url": "https://github.com/huggingface/transformers/pull/467.patch", "merged_at": 1555012404000 }

{ "login": "Akhila-Yerukola", "id": 4477323, "node_id": "MDQ6VXNlcjQ0NzczMjM=", "avatar_url": "https://avatars.githubusercontent.com/u/4477323?v=4", "gravatar_id": "", "url": "https://api.github.com/users/Akhila-Yerukola", "html_url": "https://github.com/Akhila-Yerukola", "followers_url": "https://api.github.com/users/Akhila-Yerukola/followers", "following_url": "https://api.github.com/users/Akhila-Yerukola/following{/other_user}", "gists_url": "https://api.github.com/users/Akhila-Yerukola/gists{/gist_id}", "starred_url": "https://api.github.com/users/Akhila-Yerukola/starred{/owner}{/repo}", "subscriptions_url": "https://api.github.com/users/Akhila-Yerukola/subscriptions", "organizations_url": "https://api.github.com/users/Akhila-Yerukola/orgs", "repos_url": "https://api.github.com/users/Akhila-Yerukola/repos", "events_url": "https://api.github.com/users/Akhila-Yerukola/events{/privacy}", "received_events_url": "https://api.github.com/users/Akhila-Yerukola/received_events", "type": "User", "site_admin": false }

[]

[]

[ "You are right, thanks for catching this.\r\n\r\nThis behavior is inherited from the BERT Tokenizer but the TransformerXL Tokenizer should behave differently ([this line](https://github.com/huggingface/pytorch-pretrained-BERT/blob/master/pytorch_pretrained_bert/tokenization_transfo_xl.py#L316) which split on punctuation is not present in the original tokenizer of Transformer XL [here](https://github.com/kimiyoung/transformer-xl/blob/master/tf/vocabulary.py#L25-L42)).\r\n\r\nI'll check there no other differences, add a test on this and fix this in the next release.", "@thomwolf Hello thomwolf, first of all very ty for your library. I'm reproducing transformer-xl benchmark performance using hugginface Tokenizer, but I think there is still mismatch. I just want to ask you if its solved.\r\ni would very much appreciated if you reply this :) " ]

{ "url": "https://api.github.com/repos/huggingface/transformers/issues/466/reactions", "total_count": 0, "+1": 0, "-1": 0, "laugh": 0, "hooray": 0, "confused": 0, "heart": 0, "rocket": 0, "eyes": 0 }

{ "login": "ibeltagy", "id": 2287797, "node_id": "MDQ6VXNlcjIyODc3OTc=", "avatar_url": "https://avatars.githubusercontent.com/u/2287797?v=4", "gravatar_id": "", "url": "https://api.github.com/users/ibeltagy", "html_url": "https://github.com/ibeltagy", "followers_url": "https://api.github.com/users/ibeltagy/followers", "following_url": "https://api.github.com/users/ibeltagy/following{/other_user}", "gists_url": "https://api.github.com/users/ibeltagy/gists{/gist_id}", "starred_url": "https://api.github.com/users/ibeltagy/starred{/owner}{/repo}", "subscriptions_url": "https://api.github.com/users/ibeltagy/subscriptions", "organizations_url": "https://api.github.com/users/ibeltagy/orgs", "repos_url": "https://api.github.com/users/ibeltagy/repos", "events_url": "https://api.github.com/users/ibeltagy/events{/privacy}", "received_events_url": "https://api.github.com/users/ibeltagy/received_events", "type": "User", "site_admin": false }

[ { "id": 1108649053, "node_id": "MDU6TGFiZWwxMTA4NjQ5MDUz", "url": "https://api.github.com/repos/huggingface/transformers/labels/Help%20wanted", "name": "Help wanted", "color": "008672", "default": false, "description": "Extra attention is needed, help appreciated" }, { "id": 1108649070, "node_id": "MDU6TGFiZWwxMTA4NjQ5MDcw", "url": "https://api.github.com/repos/huggingface/transformers/labels/Need%20more%20information", "name": "Need more information", "color": "d876e3", "default": false, "description": "Further information is requested" }, { "id": 1314768611, "node_id": "MDU6TGFiZWwxMzE0NzY4NjEx", "url": "https://api.github.com/repos/huggingface/transformers/labels/wontfix", "name": "wontfix", "color": "ffffff", "default": true, "description": null } ]

[]

[ "Can you try to run it with `CUDA_LAUNCH_BLOCKING=1` so we can see which exact CUDA call fails?\r\nAlso, do you have a simple way for me to try to reproduce this error?", "This issue has been automatically marked as stale because it has not had recent activity. It will be closed if no further activity occurs. Thank you for your contributions.\n" ]

{ "url": "https://api.github.com/repos/huggingface/transformers/issues/465/reactions", "total_count": 1, "+1": 1, "-1": 0, "laugh": 0, "hooray": 0, "confused": 0, "heart": 0, "rocket": 0, "eyes": 0 }

{ "login": "search4mahesh", "id": 4182331, "node_id": "MDQ6VXNlcjQxODIzMzE=", "avatar_url": "https://avatars.githubusercontent.com/u/4182331?v=4", "gravatar_id": "", "url": "https://api.github.com/users/search4mahesh", "html_url": "https://github.com/search4mahesh", "followers_url": "https://api.github.com/users/search4mahesh/followers", "following_url": "https://api.github.com/users/search4mahesh/following{/other_user}", "gists_url": "https://api.github.com/users/search4mahesh/gists{/gist_id}", "starred_url": "https://api.github.com/users/search4mahesh/starred{/owner}{/repo}", "subscriptions_url": "https://api.github.com/users/search4mahesh/subscriptions", "organizations_url": "https://api.github.com/users/search4mahesh/orgs", "repos_url": "https://api.github.com/users/search4mahesh/repos", "events_url": "https://api.github.com/users/search4mahesh/events{/privacy}", "received_events_url": "https://api.github.com/users/search4mahesh/received_events", "type": "User", "site_admin": false }

[]

[]

[ "You can get `pytorch_model.bin` and `config.json` just as indicated in the examples: https://github.com/huggingface/pytorch-pretrained-BERT/blob/master/examples/run_classifier.py#L861-L866\r\n\r\nThe vocabulary stays the same, just load the tokenizer as you did for the training (`BertTokenizer.from_pretrained(...)`)", "Thanks Thomas. I really appreciate it.\r\nOur custom text might have some new words which may not be in original vocab.txt\r\nOriginal vocab.txt contain some unused placeholders. (some 993) out of total 30522\r\nPlease let me know your thoughts on this.\r\nThanks\r\nMahesh", "Then you should go have a look at https://github.com/huggingface/pytorch-pretrained-BERT/issues/463.\r\nAnd I will close this issue in favor of #463 ;-)" ]

{ "url": "https://api.github.com/repos/huggingface/transformers/issues/464/reactions", "total_count": 0, "+1": 0, "-1": 0, "laugh": 0, "hooray": 0, "confused": 0, "heart": 0, "rocket": 0, "eyes": 0 }

{ "login": "bhoomit", "id": 1269954, "node_id": "MDQ6VXNlcjEyNjk5NTQ=", "avatar_url": "https://avatars.githubusercontent.com/u/1269954?v=4", "gravatar_id": "", "url": "https://api.github.com/users/bhoomit", "html_url": "https://github.com/bhoomit", "followers_url": "https://api.github.com/users/bhoomit/followers", "following_url": "https://api.github.com/users/bhoomit/following{/other_user}", "gists_url": "https://api.github.com/users/bhoomit/gists{/gist_id}", "starred_url": "https://api.github.com/users/bhoomit/starred{/owner}{/repo}", "subscriptions_url": "https://api.github.com/users/bhoomit/subscriptions", "organizations_url": "https://api.github.com/users/bhoomit/orgs", "repos_url": "https://api.github.com/users/bhoomit/repos", "events_url": "https://api.github.com/users/bhoomit/events{/privacy}", "received_events_url": "https://api.github.com/users/bhoomit/received_events", "type": "User", "site_admin": false }

[ { "id": 1260952223, "node_id": "MDU6TGFiZWwxMjYwOTUyMjIz", "url": "https://api.github.com/repos/huggingface/transformers/labels/Discussion", "name": "Discussion", "color": "22870e", "default": false, "description": "Discussion on a topic (keep it focused or open a new issue though)" }, { "id": 1314768611, "node_id": "MDU6TGFiZWwxMzE0NzY4NjEx", "url": "https://api.github.com/repos/huggingface/transformers/labels/wontfix", "name": "wontfix", "color": "ffffff", "default": true, "description": null } ]

[]

[ "Hi,\r\nAlso it should produce vocab.txt and bert_config.json along with pytorch_model.bin.\r\nHow you are getting those?", "We did this for [SciBERT](https://github.com/allenai/scibert), and you might find this discussion useful https://github.com/allenai/scibert/issues/29", "lm_finetuning produce pytorch_model.bin alone (and not bert_config.json)\r\nwhat do you think @Rocketknight1 ?", "```\r\nModel name '../../models/bert/' was not found in model name list (bert-base-uncased, \r\nbert-large-uncased, bert-base-cased, bert-large-cased, bert-base-multilingual-uncased,\r\nbert-base-multilingual-cased, bert-base-chinese). We assumed '../../models/bert/vocab.txt'\r\nwas a path or url but couldn't find any file associated to this path or url.\r\n```\r\n\r\nYet your fine-tuning script does not produce any such file.", "The `lm_finetuning` script assumes you're using one of the existing models in the repo, that you're fine-tuning it for a narrower domain in the same language, and that the saved pytorch_model.bin is basically just updated weights for that model - it doesn't support changes in vocab. Altering the vocab and config would probably require more extensive retraining of the model, possibly from scratch, which this repo isn't supporting yet because of the requirement for TPUs to do it quickly enough.\r\n\r\nI can contribute code if @thomwolf thinks it's relevant, but I'm not sure if or how we should be supporting this use-case right now. It might have to wait until we add from-scratch training and TPU support.", "Hi @Rocketknight1\r\nDoes this mean that the [pytorch_BERT](https://github.com/huggingface/pytorch-pretrained-BERT/) and also the [google_BERT](https://github.com/google-research/bert) implementation do not support *finetuning* with new vocabulary, sentences respectively?\r\n\r\nI would like to train a german model on a domain-specific text: the amount of german words in the multilingual model is relatively small and so I cannot access hidden states for out-of-vocabulary words even when using synonyms generated using [*FastText*](https://radimrehurek.com/gensim/models/fasttext.html), as also those synonyms are out of vocabulary. Is there any suggestion you can give me to alleviate this problem?\r\nI see that [*Issue 405*](https://github.com/huggingface/pytorch-pretrained-BERT/issues/405) has some suggestions together with [*Issue 9*](https://github.com/google-research/bert/issues/9):\r\nCan I really achieve my goal by appending my vocabulary to the end of *vocab.txt* and adjusting the *config.json* accordingly? Do I need to use the [google bert model](https://github.com/google-research/bert) and subsequently [convert_tf_checkpoint_to_pytorch.py](https://github.com/huggingface/pytorch-pretrained-BERT/blob/master/pytorch_pretrained_bert/convert_tf_checkpoint_to_pytorch.py) or can I use this repo somehow directly?", "Hi @gro1m \r\nAny luck with adding vocab to bert_pytorch?", "\r\n\r\n> Hi @gro1m\r\n> Any luck with adding vocab to bert_pytorch?\r\n\r\nI'm using https://github.com/kwonmha/bert-vocab-builder to build Vocab. Will share experience. ", "Hi, I am trying to use SciBert, the version with it's own vocab. I am wondering how to point to that vocab.txt file, and not the original. \r\n\r\nEdit\r\n\r\nFound the answer\r\n\r\nhttps://github.com/huggingface/pytorch-transformers/issues/69#issuecomment-443215315\r\n\r\nyou can just do a direct path to it", "This issue has been automatically marked as stale because it has not had recent activity. It will be closed if no further activity occurs. Thank you for your contributions.\n", "@bhoomit \r\nTo achieve BERT level results for Hinglish. would fine-tuning BERT English model with Hinglish data(approx 200 MB) could achieve good results? \r\nor it would be best to train the model from scratch in case of hinglish ? " ]

{ "url": "https://api.github.com/repos/huggingface/transformers/issues/463/reactions", "total_count": 0, "+1": 0, "-1": 0, "laugh": 0, "hooray": 0, "confused": 0, "heart": 0, "rocket": 0, "eyes": 0 }

{ "login": "8enmann", "id": 1021104, "node_id": "MDQ6VXNlcjEwMjExMDQ=", "avatar_url": "https://avatars.githubusercontent.com/u/1021104?v=4", "gravatar_id": "", "url": "https://api.github.com/users/8enmann", "html_url": "https://github.com/8enmann", "followers_url": "https://api.github.com/users/8enmann/followers", "following_url": "https://api.github.com/users/8enmann/following{/other_user}", "gists_url": "https://api.github.com/users/8enmann/gists{/gist_id}", "starred_url": "https://api.github.com/users/8enmann/starred{/owner}{/repo}", "subscriptions_url": "https://api.github.com/users/8enmann/subscriptions", "organizations_url": "https://api.github.com/users/8enmann/orgs", "repos_url": "https://api.github.com/users/8enmann/repos", "events_url": "https://api.github.com/users/8enmann/events{/privacy}", "received_events_url": "https://api.github.com/users/8enmann/received_events", "type": "User", "site_admin": false }

[]

[]

[ "Fixes #412 ", "Ok, looks good to me, thanks!" ]

{ "url": "https://api.github.com/repos/huggingface/transformers/issues/462/reactions", "total_count": 0, "+1": 0, "-1": 0, "laugh": 0, "hooray": 0, "confused": 0, "heart": 0, "rocket": 0, "eyes": 0 }

{ "url": "https://api.github.com/repos/huggingface/transformers/pulls/462", "html_url": "https://github.com/huggingface/transformers/pull/462", "diff_url": "https://github.com/huggingface/transformers/pull/462.diff", "patch_url": "https://github.com/huggingface/transformers/pull/462.patch", "merged_at": 1555012487000 }

{ "login": "meetps", "id": 6251729, "node_id": "MDQ6VXNlcjYyNTE3Mjk=", "avatar_url": "https://avatars.githubusercontent.com/u/6251729?v=4", "gravatar_id": "", "url": "https://api.github.com/users/meetps", "html_url": "https://github.com/meetps", "followers_url": "https://api.github.com/users/meetps/followers", "following_url": "https://api.github.com/users/meetps/following{/other_user}", "gists_url": "https://api.github.com/users/meetps/gists{/gist_id}", "starred_url": "https://api.github.com/users/meetps/starred{/owner}{/repo}", "subscriptions_url": "https://api.github.com/users/meetps/subscriptions", "organizations_url": "https://api.github.com/users/meetps/orgs", "repos_url": "https://api.github.com/users/meetps/repos", "events_url": "https://api.github.com/users/meetps/events{/privacy}", "received_events_url": "https://api.github.com/users/meetps/received_events", "type": "User", "site_admin": false }

[]

[]

[ "Indeed this looks like a bug in the `run_swag.py` example.\r\nWhat do you think @rodgzilla?\r\nIsn't the exclusion of the pooler parameters from optimization ([line 392 of `run_swag.py`](https://github.com/huggingface/pytorch-pretrained-BERT/blob/master/examples/run_swag.py#L392)) a typo?", "This issue has been automatically marked as stale because it has not had recent activity. It will be closed if no further activity occurs. Thank you for your contributions.\n", "@thomwolf - This surely looks like a bug, should I just send a hotfix and close it?", "Yes!", "Fixed in #675." ]

{ "url": "https://api.github.com/repos/huggingface/transformers/issues/461/reactions", "total_count": 1, "+1": 1, "-1": 0, "laugh": 0, "hooray": 0, "confused": 0, "heart": 0, "rocket": 0, "eyes": 0 }

{ "login": "prematurelyoptimized", "id": 13842527, "node_id": "MDQ6VXNlcjEzODQyNTI3", "avatar_url": "https://avatars.githubusercontent.com/u/13842527?v=4", "gravatar_id": "", "url": "https://api.github.com/users/prematurelyoptimized", "html_url": "https://github.com/prematurelyoptimized", "followers_url": "https://api.github.com/users/prematurelyoptimized/followers", "following_url": "https://api.github.com/users/prematurelyoptimized/following{/other_user}", "gists_url": "https://api.github.com/users/prematurelyoptimized/gists{/gist_id}", "starred_url": "https://api.github.com/users/prematurelyoptimized/starred{/owner}{/repo}", "subscriptions_url": "https://api.github.com/users/prematurelyoptimized/subscriptions", "organizations_url": "https://api.github.com/users/prematurelyoptimized/orgs", "repos_url": "https://api.github.com/users/prematurelyoptimized/repos", "events_url": "https://api.github.com/users/prematurelyoptimized/events{/privacy}", "received_events_url": "https://api.github.com/users/prematurelyoptimized/received_events", "type": "User", "site_admin": false }

[ { "id": 1314768611, "node_id": "MDU6TGFiZWwxMzE0NzY4NjEx", "url": "https://api.github.com/repos/huggingface/transformers/labels/wontfix", "name": "wontfix", "color": "ffffff", "default": true, "description": null } ]

[]

[ "My guess is that the inputs are either larger than the maximum input size of the model (512) or outside the vocabulary (larger than the vocabulary size).\r\n\r\nDo you think you can try to check this?", "@ananyahjha93 did the implementation of the additional GLUE tasks. Maybe he has some additional insights on this.", "@prematurelyoptimized I have not been able to re-create this error with a batch size of 8 on CoLA. I will post an update if I find something on this. In the meantime, is it possible for you to check this code on a different GPU? Also, can you post the version of PyTorch you are using? ", "My apologies for the delay, I have been out of town.\r\n\r\n@ananyahjha93 Unfortunately, I only have the one GPU for testing. print(torch.__version__) returns 1.1.0a0+be364ac\r\n\r\n@thomwolf I didn't change the tokenizer at all, so it shouldn't be an out-of-vocabulary problem and I didn't change either convert_examples_to_features or _truncate_seq_pair, so it shouldn't be a problem with the input being too long. I can, however, dig in and verify those assumptions.\r\n\r\nUnfortunately, my IT group installed a new NVidia license and now torch reports cuda being unavailable. I should still be able to check the OOV and input overflow problems, but any testing on the GPU I will have to put on hold until we get that issue sorted out.", "Ok, I am back up and running. Neither I nor my IT folks know what was causing the issue with cuda being unavailable (It was an error 35, indicating a newer runtime than the driver, but it was the same runtime and driver that were working before). In any case, upgrading to CUDA 10.1 and driver version 418.40.04 resolved the unavailability issue after rebuilding the container for the new environment. So now I can start addressing the illegal memory access issue again (sadly, it was not resolved with the driver update)\r\n\r\nI added a validation check before training to ensure that the tokenizer could convert the ids back into tokens and the length of the training samples didn't exceed the maximum length. All the checks passed, so that should rule out OOV and input overflow.\r\n\r\nI also noticed that the illegal memory access in non-deterministic. One example stack is the following:\r\n\r\nTraceback (most recent call last): | 0/1069 [00:00<?, ?it/s]\r\n File \"run_classifier2.py\", line 677, in <module>\r\n main()\r\n File \"run_classifier2.py\", line 580, in main\r\n loss = model(input_ids, segment_ids, input_mask, label_ids)\r\n File \"/opt/conda/lib/python3.6/site-packages/torch/nn/modules/module.py\", line 491, in __call__\r\n result = self.forward(*input, **kwargs)\r\n File \"/opt/conda/lib/python3.6/site-packages/pytorch_pretrained_bert/modeling.py\", line 970, in forward\r\n _, pooled_output = self.bert(input_ids, token_type_ids, attention_mask, output_all_encoded_layers=False)\r\n File \"/opt/conda/lib/python3.6/site-packages/torch/nn/modules/module.py\", line 491, in __call__\r\n result = self.forward(*input, **kwargs)\r\n File \"/opt/conda/lib/python3.6/site-packages/pytorch_pretrained_bert/modeling.py\", line 711, in forward\r\n embedding_output = self.embeddings(input_ids, token_type_ids)\r\n File \"/opt/conda/lib/python3.6/site-packages/torch/nn/modules/module.py\", line 491, in __call__\r\n result = self.forward(*input, **kwargs)\r\n File \"/opt/conda/lib/python3.6/site-packages/pytorch_pretrained_bert/modeling.py\", line 261, in forward\r\n embeddings = words_embeddings + position_embeddings + token_type_embeddings\r\nRuntimeError: CUDA error: an illegal memory access was encountered\r\n\r\nHowever, running the same command immediately afterward gives\r\n\r\nTraceback (most recent call last): | 0/1069 [00:00<?, ?it/s]\r\n File \"run_classifier2.py\", line 677, in <module>\r\n main()\r\n File \"run_classifier2.py\", line 580, in main\r\n loss = model(input_ids, segment_ids, input_mask, label_ids)\r\n File \"/opt/conda/lib/python3.6/site-packages/torch/nn/modules/module.py\", line 491, in __call__\r\n result = self.forward(*input, **kwargs)\r\n File \"/opt/conda/lib/python3.6/site-packages/pytorch_pretrained_bert/modeling.py\", line 970, in forward\r\n _, pooled_output = self.bert(input_ids, token_type_ids, attention_mask, output_all_encoded_layers=False)\r\n File \"/opt/conda/lib/python3.6/site-packages/torch/nn/modules/module.py\", line 491, in __call__\r\n result = self.forward(*input, **kwargs)\r\n File \"/opt/conda/lib/python3.6/site-packages/pytorch_pretrained_bert/modeling.py\", line 714, in forward\r\n output_all_encoded_layers=output_all_encoded_layers)\r\n File \"/opt/conda/lib/python3.6/site-packages/torch/nn/modules/module.py\", line 491, in __call__\r\n result = self.forward(*input, **kwargs)\r\n File \"/opt/conda/lib/python3.6/site-packages/pytorch_pretrained_bert/modeling.py\", line 396, in forward\r\n hidden_states = layer_module(hidden_states, attention_mask)\r\n File \"/opt/conda/lib/python3.6/site-packages/torch/nn/modules/module.py\", line 491, in __call__\r\n result = self.forward(*input, **kwargs)\r\n File \"/opt/conda/lib/python3.6/site-packages/pytorch_pretrained_bert/modeling.py\", line 382, in forward\r\n intermediate_output = self.intermediate(attention_output)\r\n File \"/opt/conda/lib/python3.6/site-packages/torch/nn/modules/module.py\", line 491, in __call__\r\n result = self.forward(*input, **kwargs)\r\n File \"/opt/conda/lib/python3.6/site-packages/pytorch_pretrained_bert/modeling.py\", line 354, in forward\r\n hidden_states = self.dense(hidden_states)\r\n File \"/opt/conda/lib/python3.6/site-packages/torch/nn/modules/module.py\", line 491, in __call__\r\n result = self.forward(*input, **kwargs)\r\n File \"/opt/conda/lib/python3.6/site-packages/torch/nn/modules/linear.py\", line 67, in forward\r\n return F.linear(input, self.weight, self.bias)\r\n File \"/opt/conda/lib/python3.6/site-packages/torch/nn/functional.py\", line 1401, in linear\r\n output = input.matmul(weight.t())\r\nRuntimeError: cublas runtime error : the GPU program failed to execute at /tmp/pip-req-build-k527kqpu/aten/src/THC/THCBlas.cu:259\r\n\r\nAnd then again,\r\n\r\nTraceback (most recent call last):\r\n File \"run_classifier2.py\", line 677, in <module>\r\n main()\r\n File \"run_classifier2.py\", line 580, in main\r\n loss = model(input_ids, segment_ids, input_mask, label_ids)\r\n File \"/opt/conda/lib/python3.6/site-packages/torch/nn/modules/module.py\", line 491, in __call__\r\n result = self.forward(*input, **kwargs)\r\n File \"/opt/conda/lib/python3.6/site-packages/pytorch_pretrained_bert/modeling.py\", line 970, in forward\r\n _, pooled_output = self.bert(input_ids, token_type_ids, attention_mask, output_all_encoded_layers=False)\r\n File \"/opt/conda/lib/python3.6/site-packages/torch/nn/modules/module.py\", line 491, in __call__\r\n result = self.forward(*input, **kwargs)\r\n File \"/opt/conda/lib/python3.6/site-packages/pytorch_pretrained_bert/modeling.py\", line 714, in forward\r\n output_all_encoded_layers=output_all_encoded_layers)\r\n File \"/opt/conda/lib/python3.6/site-packages/torch/nn/modules/module.py\", line 491, in __call__\r\n result = self.forward(*input, **kwargs)\r\n File \"/opt/conda/lib/python3.6/site-packages/pytorch_pretrained_bert/modeling.py\", line 396, in forward\r\n hidden_states = layer_module(hidden_states, attention_mask)\r\n File \"/opt/conda/lib/python3.6/site-packages/torch/nn/modules/module.py\", line 491, in __call__\r\n result = self.forward(*input, **kwargs)\r\n File \"/opt/conda/lib/python3.6/site-packages/pytorch_pretrained_bert/modeling.py\", line 381, in forward\r\n attention_output = self.attention(hidden_states, attention_mask)\r\n File \"/opt/conda/lib/python3.6/site-packages/torch/nn/modules/module.py\", line 491, in __call__\r\n result = self.forward(*input, **kwargs)\r\n File \"/opt/conda/lib/python3.6/site-packages/pytorch_pretrained_bert/modeling.py\", line 339, in forward\r\n self_output = self.self(input_tensor, attention_mask)\r\n File \"/opt/conda/lib/python3.6/site-packages/torch/nn/modules/module.py\", line 491, in __call__\r\n result = self.forward(*input, **kwargs)\r\n File \"/opt/conda/lib/python3.6/site-packages/pytorch_pretrained_bert/modeling.py\", line 305, in forward\r\n attention_probs = nn.Softmax(dim=-1)(attention_scores)\r\n File \"/opt/conda/lib/python3.6/site-packages/torch/nn/modules/module.py\", line 491, in __call__\r\n result = self.forward(*input, **kwargs)\r\n File \"/opt/conda/lib/python3.6/site-packages/torch/nn/modules/activation.py\", line 877, in forward\r\n return F.softmax(input, self.dim, _stacklevel=5)\r\n File \"/opt/conda/lib/python3.6/site-packages/torch/nn/functional.py\", line 1256, in softmax\r\n ret = input.softmax(dim)\r\nRuntimeError: cuda runtime error (700) : an illegal memory access was encountered at /tmp/pip-req-build-k527kqpu/aten/src/ATen/native/cuda/SoftMax.cu:545\r\n\r\nThis feels like an OOM error, but watching nvidia-smi, the memory usage peaks at only about 800MiB. This doesn't change whether my batch size is 2, 4, 8, or 32, so it looks like the batch size is not the culprit either. The non-determinism could be caused by the kernel running asynchronously, but that shouldn't happen since I'm running with CUDA_LAUNCH_BLOCKING=1.\r\n\r\nLooking at the pytorch source at the lines in the stack traces hasn't been illuminating, so are there some recommended debugging steps at this point?", "This issue has been automatically marked as stale because it has not had recent activity. It will be closed if no further activity occurs. Thank you for your contributions.\n" ]

{ "url": "https://api.github.com/repos/huggingface/transformers/issues/460/reactions", "total_count": 0, "+1": 0, "-1": 0, "laugh": 0, "hooray": 0, "confused": 0, "heart": 0, "rocket": 0, "eyes": 0 }

{ "login": "geekboood", "id": 23272969, "node_id": "MDQ6VXNlcjIzMjcyOTY5", "avatar_url": "https://avatars.githubusercontent.com/u/23272969?v=4", "gravatar_id": "", "url": "https://api.github.com/users/geekboood", "html_url": "https://github.com/geekboood", "followers_url": "https://api.github.com/users/geekboood/followers", "following_url": "https://api.github.com/users/geekboood/following{/other_user}", "gists_url": "https://api.github.com/users/geekboood/gists{/gist_id}", "starred_url": "https://api.github.com/users/geekboood/starred{/owner}{/repo}", "subscriptions_url": "https://api.github.com/users/geekboood/subscriptions", "organizations_url": "https://api.github.com/users/geekboood/orgs", "repos_url": "https://api.github.com/users/geekboood/repos", "events_url": "https://api.github.com/users/geekboood/events{/privacy}", "received_events_url": "https://api.github.com/users/geekboood/received_events", "type": "User", "site_admin": false }

[]

[]

[]

{ "url": "https://api.github.com/repos/huggingface/transformers/issues/459/reactions", "total_count": 0, "+1": 0, "-1": 0, "laugh": 0, "hooray": 0, "confused": 0, "heart": 0, "rocket": 0, "eyes": 0 }

{ "login": "mboyanov", "id": 1610015, "node_id": "MDQ6VXNlcjE2MTAwMTU=", "avatar_url": "https://avatars.githubusercontent.com/u/1610015?v=4", "gravatar_id": "", "url": "https://api.github.com/users/mboyanov", "html_url": "https://github.com/mboyanov", "followers_url": "https://api.github.com/users/mboyanov/followers", "following_url": "https://api.github.com/users/mboyanov/following{/other_user}", "gists_url": "https://api.github.com/users/mboyanov/gists{/gist_id}", "starred_url": "https://api.github.com/users/mboyanov/starred{/owner}{/repo}", "subscriptions_url": "https://api.github.com/users/mboyanov/subscriptions", "organizations_url": "https://api.github.com/users/mboyanov/orgs", "repos_url": "https://api.github.com/users/mboyanov/repos", "events_url": "https://api.github.com/users/mboyanov/events{/privacy}", "received_events_url": "https://api.github.com/users/mboyanov/received_events", "type": "User", "site_admin": false }

[ { "id": 1260952223, "node_id": "MDU6TGFiZWwxMjYwOTUyMjIz", "url": "https://api.github.com/repos/huggingface/transformers/labels/Discussion", "name": "Discussion", "color": "22870e", "default": false, "description": "Discussion on a topic (keep it focused or open a new issue though)" } ]

[]

[ "I understand the issue but I'm not sure this would be very easy to implement as we would like to keep the model and tokenizer separated one from the other.\r\n\r\nDo you have a solution in mind?\r\n\r\nOtherwise, I'll guess people will have to continue to read the paper before using the model... 😉", "I totally understand the need for separation between tokenizer and model. \r\nBut, the specific `BertForSequenceClassification` does have the assumption and expects the input to be in a specific form. Even the docs mention \"look at the preprocessing logic\". \r\nI'd suggest extending the documentation for the model - sth like https://github.com/huggingface/pytorch-pretrained-BERT/pull/480" ]

{ "url": "https://api.github.com/repos/huggingface/transformers/issues/458/reactions", "total_count": 0, "+1": 0, "-1": 0, "laugh": 0, "hooray": 0, "confused": 0, "heart": 0, "rocket": 0, "eyes": 0 }

{ "login": "sheetalsh456", "id": 12454534, "node_id": "MDQ6VXNlcjEyNDU0NTM0", "avatar_url": "https://avatars.githubusercontent.com/u/12454534?v=4", "gravatar_id": "", "url": "https://api.github.com/users/sheetalsh456", "html_url": "https://github.com/sheetalsh456", "followers_url": "https://api.github.com/users/sheetalsh456/followers", "following_url": "https://api.github.com/users/sheetalsh456/following{/other_user}", "gists_url": "https://api.github.com/users/sheetalsh456/gists{/gist_id}", "starred_url": "https://api.github.com/users/sheetalsh456/starred{/owner}{/repo}", "subscriptions_url": "https://api.github.com/users/sheetalsh456/subscriptions", "organizations_url": "https://api.github.com/users/sheetalsh456/orgs", "repos_url": "https://api.github.com/users/sheetalsh456/repos", "events_url": "https://api.github.com/users/sheetalsh456/events{/privacy}", "received_events_url": "https://api.github.com/users/sheetalsh456/received_events", "type": "User", "site_admin": false }

[ { "id": 1260952223, "node_id": "MDU6TGFiZWwxMjYwOTUyMjIz", "url": "https://api.github.com/repos/huggingface/transformers/labels/Discussion", "name": "Discussion", "color": "22870e", "default": false, "description": "Discussion on a topic (keep it focused or open a new issue though)" }, { "id": 1314768611, "node_id": "MDU6TGFiZWwxMzE0NzY4NjEx", "url": "https://api.github.com/repos/huggingface/transformers/labels/wontfix", "name": "wontfix", "color": "ffffff", "default": true, "description": null } ]

[]

[ "Have you tried the solutions discussed in the other issues on this topic:\r\n- https://github.com/huggingface/pytorch-pretrained-BERT/issues/312\r\n- https://github.com/huggingface/pytorch-pretrained-BERT/issues/239", "This issue has been automatically marked as stale because it has not had recent activity. It will be closed if no further activity occurs. Thank you for your contributions.\n", "> Have you tried the solutions discussed in the other issues on this topic:\r\n> \r\n> * #312\r\n> * #239\r\n\r\nHi @thomwolf ,\r\n\r\nI followed the instructions [here](https://github.com/BaderLab/saber/issues/135#issuecomment-494455933) to convert the checkpoint and then placing the files (pytorch_model.bin, bert_config.json, and vocab.txt) in one folder to compress it. I did not need to ignore any weights as mentioned in any solutions you mentioned above ( #312 or #239 )\r\n\r\nI copied the compressed folder to the home folder of 'pytorch-transformers'. Then I ran the following command from there itself to run the example code (`'examples/run_glue.py'`) on my data.\r\n\r\nThen, I am trying to run the example code given by running the following command,\r\n\r\n`python ./examples/run_glue.py \\ --model_type bert \\ --model_name_or_path biobert.gz \\ --task_name=sts-b \\ --do_train \\ --do_eval \\ --do_lower_case \\ --data_dir=$DIR \\ --max_seq_length 128 \\ --per_gpu_eval_batch_size=8 \\ --per_gpu_train_batch_size=8`\r\n\r\nBut, I get the same error as mentioned in the main discussion:\r\n\r\n`UnicodeDecodeError: 'utf-8' codec can't decode byte 0x8b in position 1: invalid start byte`\r\n\r\nat this location:\r\n\r\n`File \"./examples/run_glue.py\", line 424, in main config = config_class.from_pretrained(args.config_name if args.config_name else args.model_name_or_path, num_labels=num_labels, finetuning_task=args.task_name)`\r\n\r\nCan you please tell what to change.", "Try to pass the extracted folder of your converted bioBERT model to the `--model_name_or_path` :)\r\n\r\nHere's a short example:\r\n\r\n* Download the *BioBERT v1.1 (+ PubMed 1M)* model (or any other model) from the [bioBERT repo](https://github.com/naver/biobert-pretrained)\r\n* Extract the downloaded file, e.g. with `tar -xzf biobert_v1.1_pubmed.tar.gz`\r\n* Convert the bioBERT model TensorFlow checkpoint to a PyTorch and PyTorch-Transformers compatible one: `pytorch_transformers bert biobert_v1.1_pubmed/model.ckpt-1000000 biobert_v1.1_pubmed/bert_config.json biobert_v1.1_pubmed/pytorch_model.bin`\r\n* Move config `mv biobert_v1.1_pubmed/bert_config.json biobert_v1.1_pubmed/config.json`\r\n\r\nThen pass the folder name to the `--model_name_or_path` argument. You can run this simple script to check, if everything works:\r\n\r\n```python\r\nfrom pytorch_transformers import BertModel\r\nmodel = BertModel.from_pretrained('biobert_v1.1_pubmed')\r\n```", "How we load a tensor as a pretrained model in bert", "@stefan-it As per new `transformers-cli` third command would change as follows:\r\n\r\n```bat\r\ntransformers-cli convert --model_type bert \\\r\n--tf_checkpoint biobert_v1.1_pubmed/model.ckpt-1000000 \\\r\n--config biobert_v1.1_pubmed/bert_config.json \\\r\n--pytorch_dump_output biobert_v1.1_pubmed/pytorch_model.bin\r\n```", "Hello!\r\nJust to complement the @stefan-it instructions in step number 3, it works for me the following code:\r\n\r\n`import os`\r\n`from pytorch_pretrained_bert.convert_tf_checkpoint_to_pytorch import convert_tf_checkpoint_to_pytorch`\r\n`path_bin = 'my_directory/pytorch_model.bin' `\r\n`path_bert = 'my_bert_directory/' `\r\n\r\n`if (not os.path.exists(path_bin)): `\r\n`\tconvert_tf_checkpoint_to_pytorch( `\r\n`\t\tpath_bert + \"biobert_model.ckpt\", `\r\n`\t\tpath_bert + \"bert_config.json\", `\r\n`\t\tpath_bert + \"pytorch_model.bin\" `\r\n`\t\t)`\r\n\r\nMy folder (biobert_v1.0_pmc) was originally 5 files:\r\n3 Tensorflow checkpoint files\r\nA vocab file\r\nA config file\r\n", "> * Move config `mv biobert_v1.1_pubmed/bert_config.json biobert_v1.1_pubmed/config.json`\r\n\r\nTnank you very much! It did help me!\r\n", "Thank you every one, this works fine now!", "Dear all,\r\nThank you very much for the suggestions on how to prepare the model to be used with hugginfaces. I am trying to use huggingfaces BertTokenizer to perform NER on biomedical data with the pre-trained weights. \r\nIt seems to work fine until then point when I want to map annotated tokens to entity labels. I have token ids and prediction ids, but I cannot figure out how/where to get label_list to follow an example of mapping from https://huggingface.co./transformers/usage.html#named-entity-recognition \r\nThank you very much for any help you can provide!\r\nMaria", "Dear all,\r\nI am a newbie and I don not have so much experience. Does anyone have a full tutorial or code for a regression task, Please share with me! I would greatly appreciate it! Thank you.", "@stefan-it @nipunsadvilkar Thank you for your solutions." ]

{ "url": "https://api.github.com/repos/huggingface/transformers/issues/457/reactions", "total_count": 0, "+1": 0, "-1": 0, "laugh": 0, "hooray": 0, "confused": 0, "heart": 0, "rocket": 0, "eyes": 0 }

{ "login": "ZhaofengWu", "id": 11954789, "node_id": "MDQ6VXNlcjExOTU0Nzg5", "avatar_url": "https://avatars.githubusercontent.com/u/11954789?v=4", "gravatar_id": "", "url": "https://api.github.com/users/ZhaofengWu", "html_url": "https://github.com/ZhaofengWu", "followers_url": "https://api.github.com/users/ZhaofengWu/followers", "following_url": "https://api.github.com/users/ZhaofengWu/following{/other_user}", "gists_url": "https://api.github.com/users/ZhaofengWu/gists{/gist_id}", "starred_url": "https://api.github.com/users/ZhaofengWu/starred{/owner}{/repo}", "subscriptions_url": "https://api.github.com/users/ZhaofengWu/subscriptions", "organizations_url": "https://api.github.com/users/ZhaofengWu/orgs", "repos_url": "https://api.github.com/users/ZhaofengWu/repos", "events_url": "https://api.github.com/users/ZhaofengWu/events{/privacy}", "received_events_url": "https://api.github.com/users/ZhaofengWu/received_events", "type": "User", "site_admin": false }

[]

[]

[ "Then you're doomed for this answer. There are a possibility to do a sliding window approach but we didn't implemented it in the examples of pytorch-pretrained-bert.\r\nCheck this issue (and the linked TensorFlow issue) for a discussion on this: https://github.com/huggingface/pytorch-pretrained-BERT/issues/89" ]

{ "url": "https://api.github.com/repos/huggingface/transformers/issues/456/reactions", "total_count": 0, "+1": 0, "-1": 0, "laugh": 0, "hooray": 0, "confused": 0, "heart": 0, "rocket": 0, "eyes": 0 }

{ "login": "ZhaofengWu", "id": 11954789, "node_id": "MDQ6VXNlcjExOTU0Nzg5", "avatar_url": "https://avatars.githubusercontent.com/u/11954789?v=4", "gravatar_id": "", "url": "https://api.github.com/users/ZhaofengWu", "html_url": "https://github.com/ZhaofengWu", "followers_url": "https://api.github.com/users/ZhaofengWu/followers", "following_url": "https://api.github.com/users/ZhaofengWu/following{/other_user}", "gists_url": "https://api.github.com/users/ZhaofengWu/gists{/gist_id}", "starred_url": "https://api.github.com/users/ZhaofengWu/starred{/owner}{/repo}", "subscriptions_url": "https://api.github.com/users/ZhaofengWu/subscriptions", "organizations_url": "https://api.github.com/users/ZhaofengWu/orgs", "repos_url": "https://api.github.com/users/ZhaofengWu/repos", "events_url": "https://api.github.com/users/ZhaofengWu/events{/privacy}", "received_events_url": "https://api.github.com/users/ZhaofengWu/received_events", "type": "User", "site_admin": false }

[ { "id": 1260952223, "node_id": "MDU6TGFiZWwxMjYwOTUyMjIz", "url": "https://api.github.com/repos/huggingface/transformers/labels/Discussion", "name": "Discussion", "color": "22870e", "default": false, "description": "Discussion on a topic (keep it focused or open a new issue though)" } ]

[]

[ "Not really, I guess we could remove this restriction and just let people point to an arbitrary folder (as we do in the other scripts).\r\nWhat do you think @Rocketknight1?", "No, that's entirely my bad. We should allow an arbitrary folder!", "Should I just remove the `choices` argument and allow any string there?", "Yes, I would go for that!", "Actually, if you have push access to the repo, do you want to just make that one-line change? I'd have to fork and submit a PR, which seems a bit unnecessary.", "Indeed, I'll take care of it. I'm on the repo now anyway.", "is this change applicable and working with run_lm_finetuning.py in 2.x?" ]

{ "url": "https://api.github.com/repos/huggingface/transformers/issues/455/reactions", "total_count": 0, "+1": 0, "-1": 0, "laugh": 0, "hooray": 0, "confused": 0, "heart": 0, "rocket": 0, "eyes": 0 }

{ "login": "omerarshad", "id": 16164105, "node_id": "MDQ6VXNlcjE2MTY0MTA1", "avatar_url": "https://avatars.githubusercontent.com/u/16164105?v=4", "gravatar_id": "", "url": "https://api.github.com/users/omerarshad", "html_url": "https://github.com/omerarshad", "followers_url": "https://api.github.com/users/omerarshad/followers", "following_url": "https://api.github.com/users/omerarshad/following{/other_user}", "gists_url": "https://api.github.com/users/omerarshad/gists{/gist_id}", "starred_url": "https://api.github.com/users/omerarshad/starred{/owner}{/repo}", "subscriptions_url": "https://api.github.com/users/omerarshad/subscriptions", "organizations_url": "https://api.github.com/users/omerarshad/orgs", "repos_url": "https://api.github.com/users/omerarshad/repos", "events_url": "https://api.github.com/users/omerarshad/events{/privacy}", "received_events_url": "https://api.github.com/users/omerarshad/received_events", "type": "User", "site_admin": false }

[]

[]

[ "Hi, yes." ]

{ "url": "https://api.github.com/repos/huggingface/transformers/issues/454/reactions", "total_count": 0, "+1": 0, "-1": 0, "laugh": 0, "hooray": 0, "confused": 0, "heart": 0, "rocket": 0, "eyes": 0 }

{ "login": "JiahangOK", "id": 40447174, "node_id": "MDQ6VXNlcjQwNDQ3MTc0", "avatar_url": "https://avatars.githubusercontent.com/u/40447174?v=4", "gravatar_id": "", "url": "https://api.github.com/users/JiahangOK", "html_url": "https://github.com/JiahangOK", "followers_url": "https://api.github.com/users/JiahangOK/followers", "following_url": "https://api.github.com/users/JiahangOK/following{/other_user}", "gists_url": "https://api.github.com/users/JiahangOK/gists{/gist_id}", "starred_url": "https://api.github.com/users/JiahangOK/starred{/owner}{/repo}", "subscriptions_url": "https://api.github.com/users/JiahangOK/subscriptions", "organizations_url": "https://api.github.com/users/JiahangOK/orgs", "repos_url": "https://api.github.com/users/JiahangOK/repos", "events_url": "https://api.github.com/users/JiahangOK/events{/privacy}", "received_events_url": "https://api.github.com/users/JiahangOK/received_events", "type": "User", "site_admin": false }

[]

[]

[ "Hi, it means that the computation graphs of the Tensorflow and PyTorch versions are identical." ]

{ "url": "https://api.github.com/repos/huggingface/transformers/issues/453/reactions", "total_count": 0, "+1": 0, "-1": 0, "laugh": 0, "hooray": 0, "confused": 0, "heart": 0, "rocket": 0, "eyes": 0 }

{ "login": "yanneyanne", "id": 15113131, "node_id": "MDQ6VXNlcjE1MTEzMTMx", "avatar_url": "https://avatars.githubusercontent.com/u/15113131?v=4", "gravatar_id": "", "url": "https://api.github.com/users/yanneyanne", "html_url": "https://github.com/yanneyanne", "followers_url": "https://api.github.com/users/yanneyanne/followers", "following_url": "https://api.github.com/users/yanneyanne/following{/other_user}", "gists_url": "https://api.github.com/users/yanneyanne/gists{/gist_id}", "starred_url": "https://api.github.com/users/yanneyanne/starred{/owner}{/repo}", "subscriptions_url": "https://api.github.com/users/yanneyanne/subscriptions", "organizations_url": "https://api.github.com/users/yanneyanne/orgs", "repos_url": "https://api.github.com/users/yanneyanne/repos", "events_url": "https://api.github.com/users/yanneyanne/events{/privacy}", "received_events_url": "https://api.github.com/users/yanneyanne/received_events", "type": "User", "site_admin": false }

[ { "id": 1260952223, "node_id": "MDU6TGFiZWwxMjYwOTUyMjIz", "url": "https://api.github.com/repos/huggingface/transformers/labels/Discussion", "name": "Discussion", "color": "22870e", "default": false, "description": "Discussion on a topic (keep it focused or open a new issue though)" }, { "id": 1314768611, "node_id": "MDU6TGFiZWwxMzE0NzY4NjEx", "url": "https://api.github.com/repos/huggingface/transformers/labels/wontfix", "name": "wontfix", "color": "ffffff", "default": true, "description": null } ]

[]

[ "Hi, I wrote that script! That's a tricky issue, though - what behaviour do you expect when your data is one long text? \r\n\r\nIn the original BERT repo, they used the document breaks to control sampling for the NextSentence task - 'random' next sentences were selected from a different document. I'm not sure there's a \"general\" solution to the case when all the data is a single document, though - people whose data is all one contiguous document will probably have different expectations for how they want that data to be handled (e.g. is it okay to select a 'random' sentence 10 sentences away, or should it be much further? Are 'nearby' sentences too similar?)\r\n\r\nI suppose we could add code to support that with a controllable parameter to stop it sampling random sentences that are too close to the actual sentence, but it might be easier just to throw an error if the data is only one single document and request that the user break it into 'documents' based on whatever structure is in the data, like chapter or page breaks, or in the worst case just add a break every 1,000 lines or something.\r\n\r\nYou're totally right about the f-strings, though. I'll look into updating the README!", "I'm assuming that the script should just grab any random sentence in the same (only) document. I could see how that particular behavior might not be desired however.\r\n\r\nThe error-throwing idea is good. I ran the script on my mammoth document, only to see empty output after 9 hours (my own fault obviously, but an exception might help similarly sloppy people).", "Added PR #478 to address this. Unfortunately, it probably still won't resolve your 9 hour problem - the script will only realize there was only one document in the corpus after it's finished reading and tokenizing the whole thing! Still, at least people will get an error message properly explaining the issue and suggesting fixes now.", "This issue has been automatically marked as stale because it has not had recent activity. It will be closed if no further activity occurs. Thank you for your contributions.\n" ]

{ "url": "https://api.github.com/repos/huggingface/transformers/issues/452/reactions", "total_count": 0, "+1": 0, "-1": 0, "laugh": 0, "hooray": 0, "confused": 0, "heart": 0, "rocket": 0, "eyes": 0 }

{ "login": "yangliu2", "id": 3577730, "node_id": "MDQ6VXNlcjM1Nzc3MzA=", "avatar_url": "https://avatars.githubusercontent.com/u/3577730?v=4", "gravatar_id": "", "url": "https://api.github.com/users/yangliu2", "html_url": "https://github.com/yangliu2", "followers_url": "https://api.github.com/users/yangliu2/followers", "following_url": "https://api.github.com/users/yangliu2/following{/other_user}", "gists_url": "https://api.github.com/users/yangliu2/gists{/gist_id}", "starred_url": "https://api.github.com/users/yangliu2/starred{/owner}{/repo}", "subscriptions_url": "https://api.github.com/users/yangliu2/subscriptions", "organizations_url": "https://api.github.com/users/yangliu2/orgs", "repos_url": "https://api.github.com/users/yangliu2/repos", "events_url": "https://api.github.com/users/yangliu2/events{/privacy}", "received_events_url": "https://api.github.com/users/yangliu2/received_events", "type": "User", "site_admin": false }

[ { "id": 1314768611, "node_id": "MDU6TGFiZWwxMzE0NzY4NjEx", "url": "https://api.github.com/repos/huggingface/transformers/labels/wontfix", "name": "wontfix", "color": "ffffff", "default": true, "description": null } ]

[]

[ "It is likely due to the script is not able to find the vocabulary file. So you should download the vocab first and copy it over. Then when you load the tokenizer you need to specify the path to the vocab. So if you vocab is in \"/tmp/transformer_xl/\" you do:\r\n```python\r\ntokenizer = TransfoXLTokenizer.from_pretrained('/tmp/transformer_xl')\r\n```", "This works, just not what I expected. I copied over everything in .pytorch_pretrained_bert and though it would load without parameters. Now I have a bunch of file named like this, I have to figure out which model it belongs.\r\n\"12642ff7d0279757d8356bfd86a729d9697018a0c93ad042de1d0d2cc17fd57b.e9704971f27275ec067a00a67e6a5f0b05b4306b3f714a96e9f763d8fb612671\"", "I will add a section in the readme detailing how to load a model from drive.\r\nBasically, you can just download the models and vocabulary from our S3 following the links at the top of each file (`modeling_transfo_xl.py` and `tokenization_transfo_xl.py` for Transformer-XL) and put them in one directory with the filename also indicated at the top of each file.\r\n\r\nHere is the process in your case:\r\n```bash\r\nmkdir model\r\ncd model\r\nwget -O pytorch_model.bin https://s3.amazonaws.com/models.huggingface.co/bert/transfo-xl-wt103-pytorch_model.bin\r\nwget -O config.json https://s3.amazonaws.com/models.huggingface.co/bert/transfo-xl-wt103-config.json\r\nwget -O vocab.bin https://s3.amazonaws.com/models.huggingface.co/bert/transfo-xl-wt103-vocab.bin\r\n# optional, only if you run the evaluation script run_transfo_xl.py which uses the pre-processed wt103 corpus:\r\n# wget -O corpus.bin https://s3.amazonaws.com/models.huggingface.co/bert/transfo-xl-wt103-corpus.bin\r\n```\r\n\r\nNow just load the model and tokenizer by pointing to this directory:\r\n```python\r\ntokenizer = TransfoXLTokenizer.from_pretrained('./model/')\r\nmodel = TransfoXLModel.from_pretrained('./model/')\r\n```\r\n", "I'll see if I can relax the requirement on the internet connection in the next release.", "The network connection check has been relaxed in the now merged #500.\r\nIt will be included in the next PyPI release (probably next week).\r\nIn the meantime you can install from `master`.", "I have a similar issue with the BERT multilingual cased model: \r\n```\r\nERROR - pytorch_pretrained_bert.modeling - Model name 'bert-base-multilingual-cased' was not found in model name list (bert-base-uncased, bert-large-uncased, bert-base-cased, bert-large-cased, bert-base-multilingual-uncased, bert-base-multilingual-cased, bert-base-chinese). \r\nWe assumed 'https://s3.amazonaws.com/models.huggingface.co/bert/bert-base-multilingual-cased.tar.gz' was a path or url but couldn't find any file associated to this path or url. \r\n```\r\nThen I tried to execute the following code block in my Jupyter notebook:\r\n```python\r\nimport requests\r\nurl = \"https://s3.amazonaws.com/models.huggingface.co/bert/bert-base-multilingual-cased.tar.gz\"\r\nresponse = requests.get(url, allow_redirects=True, verify=False)\r\n```\r\nI had to set verify to false, because otherwise I get an SSL certificate error. But even now, this does not work, because the site is blocked due to our company security settings, i.e. `response.status_code` returns 403.\r\n\r\nIs there a possibility that you might publish the file in your github repo or that we could load the model from somewhere else?", "Strange, I can't reproduce this.\r\nI've checked again that every model is public on our S3.\r\nCan you try again?", "I retried, once using the google tensorflow hub address and once with the Amazon S3 address for the BERT model.\r\nI specified the proxy information like this:\r\n```\r\nproxyDict = { \"http\" : \"http://<proxy-user>:<proxy-password>@<proxy-domain>\",\r\n \"https\" : \"http://<proxy-user>:<proxy-password>@<proxy-domain>\"}\r\n```\r\nwith our company-specific settings for proxy-user, proxy-password and proxy-domain. \r\n\r\nThen I executed the following code:\r\n```python\r\nimport requests\r\nurl_google = \"https://tfhub.dev/google/bert_multi_cased_L-12_H-768_A-12/1\"\r\nurl_amazons3 = \"https://s3.amazonaws.com/models.huggingface.co/bert/bert-base-multilingual-cased.tar.gz\"\r\n\r\nresponse_1 = requests.get(url_google, allow_redirects=True, verify=False, proxies = proxyDict)\r\nresponse_2 = requests.get(url_amazons3, allow_redirects=True, verify=False, proxies = proxyDict)\r\n\r\nprint(\"response status code for google address: {}\".format(response_1.status_code))\r\nprint(\"response status code for amazon s3 address: {}\".format(response_2.status_code))\r\n```\r\n\r\nand this is what I got:\r\n```\r\nresponse status code for google address: 200\r\nresponse status code for amazon s3 address: 403\r\n```\r\n\r\nSo unfortunately, it does not seem to work out for me. I might use the convert function you provided, but it would be nicer to be able to load the model directly from the S3.", "Is it only for this model (`bert-base-multilingual-cased`) or are you blocked from accessing all the pretrained models and tokenizers?", "I am blocked from accessing all the pretrained models.\r\nI tested it by looping through the values of PRETRAINED_MODEL_ARCHIVE_MAP dictionary and all requests return the status code 403.", "I haven't tried it yet but maybe torch hub could help (#506)\r\n\r\nCan you try to update to PyTorch 1.1.0 (to get torch.hub) and test this:\r\n```python\r\nimport torch\r\ntokenizer = torch.hub.load('ailzhang/pytorch-pretrained-BERT:hubconf', 'bertTokenizer', 'bert-base-cased', do_basic_tokenize=False, force_reload=False)\r\n```", "Well, it does not seem to work.\r\nI had to add \r\n```python\r\nimport urllib\r\nproxy_support = urllib.request.ProxyHandler({ \"http\" : \"http://<proxy-user>:<proxy-password>@<proxy-domain>\",\r\n\"https\" : \"http://<proxy-user>:<proxy-password>@<proxy-domain>\"})\r\nopener = urllib.request.build_opener(proxy_support)\r\nurllib.request.install_opener(opener)\r\n```\r\ninto ~/my-virtual-env/lib/site-packages/python3.6/torch/hub.py with my-virtual-env being my pip virtual environment.\r\nThen executing the command you suggested prints the following to the console:\r\n```\r\nDownloading: \"https://github.com/ailzhang/pytorch-pretrained-BERT/archive/hubconf.zip\" to /home/U118693/.cache/torch/hub/hubconf.zip\r\nThe pre-trained model you are loading is a cased model but you have not set `do_lower_case` to False. We are setting `do_lower_case=False` for you but you may want to check this behavior.\r\nModel name 'bert-base-cased' was not found in model name list (bert-base-uncased, bert-large-uncased, bert-base-cased, bert-large-cased, bert-base-multilingual-uncased, bert-base-multilingual-cased, bert-base-chinese). We assumed 'https://s3.amazonaws.com/models.huggingface.co/bert/bert-base-cased-vocab.txt' was a path or url but couldn't find any file associated to this path or url.\r\n```\r\nThere is now a folder ailzhang_pytorch-pretrained-BERT_hubconf in the /home/U118693/.cache/torch/hub/ directory, but there still seems to be issues in finding that bert-cased-vocab.txt file.", "This issue has been automatically marked as stale because it has not had recent activity. It will be closed if no further activity occurs. Thank you for your contributions.\n" ]

{ "url": "https://api.github.com/repos/huggingface/transformers/issues/451/reactions", "total_count": 0, "+1": 0, "-1": 0, "laugh": 0, "hooray": 0, "confused": 0, "heart": 0, "rocket": 0, "eyes": 0 }

{ "login": "BramVanroy", "id": 2779410, "node_id": "MDQ6VXNlcjI3Nzk0MTA=", "avatar_url": "https://avatars.githubusercontent.com/u/2779410?v=4", "gravatar_id": "", "url": "https://api.github.com/users/BramVanroy", "html_url": "https://github.com/BramVanroy", "followers_url": "https://api.github.com/users/BramVanroy/followers", "following_url": "https://api.github.com/users/BramVanroy/following{/other_user}", "gists_url": "https://api.github.com/users/BramVanroy/gists{/gist_id}", "starred_url": "https://api.github.com/users/BramVanroy/starred{/owner}{/repo}", "subscriptions_url": "https://api.github.com/users/BramVanroy/subscriptions", "organizations_url": "https://api.github.com/users/BramVanroy/orgs", "repos_url": "https://api.github.com/users/BramVanroy/repos", "events_url": "https://api.github.com/users/BramVanroy/events{/privacy}", "received_events_url": "https://api.github.com/users/BramVanroy/received_events", "type": "User", "site_admin": false }

[ { "id": 1260952223, "node_id": "MDU6TGFiZWwxMjYwOTUyMjIz", "url": "https://api.github.com/repos/huggingface/transformers/labels/Discussion", "name": "Discussion", "color": "22870e", "default": false, "description": "Discussion on a topic (keep it focused or open a new issue though)" }, { "id": 1314768611, "node_id": "MDU6TGFiZWwxMzE0NzY4NjEx", "url": "https://api.github.com/repos/huggingface/transformers/labels/wontfix", "name": "wontfix", "color": "ffffff", "default": true, "description": null } ]

[]

[ "Maybe a good introduction on the topic are the writings of @sebastianruder:\r\n- http://ruder.io/transfer-learning/\r\n- https://thegradient.pub/nlp-imagenet/\r\n- the ULMFiT paper: http://nlp.fast.ai/classification/2018/05/15/introducting-ulmfit.html\r\n\r\nRegarding your specific question of training a Bert model on a new language, the computing requirements for training BERT are very high. You would probably be better with a version of ULMFiT. There is a huge multilingual initiative on the fast.ai forum that you can check out. One entry point is [here](https://forums.fast.ai/t/language-model-zoo-gorilla/14623).", "This issue has been automatically marked as stale because it has not had recent activity. It will be closed if no further activity occurs. Thank you for your contributions.\n" ]

{ "url": "https://api.github.com/repos/huggingface/transformers/issues/450/reactions", "total_count": 0, "+1": 0, "-1": 0, "laugh": 0, "hooray": 0, "confused": 0, "heart": 0, "rocket": 0, "eyes": 0 }

{ "login": "raheja", "id": 9508410, "node_id": "MDQ6VXNlcjk1MDg0MTA=", "avatar_url": "https://avatars.githubusercontent.com/u/9508410?v=4", "gravatar_id": "", "url": "https://api.github.com/users/raheja", "html_url": "https://github.com/raheja", "followers_url": "https://api.github.com/users/raheja/followers", "following_url": "https://api.github.com/users/raheja/following{/other_user}", "gists_url": "https://api.github.com/users/raheja/gists{/gist_id}", "starred_url": "https://api.github.com/users/raheja/starred{/owner}{/repo}", "subscriptions_url": "https://api.github.com/users/raheja/subscriptions", "organizations_url": "https://api.github.com/users/raheja/orgs", "repos_url": "https://api.github.com/users/raheja/repos", "events_url": "https://api.github.com/users/raheja/events{/privacy}", "received_events_url": "https://api.github.com/users/raheja/received_events", "type": "User", "site_admin": false }

[ { "id": 1108649053, "node_id": "MDU6TGFiZWwxMTA4NjQ5MDUz", "url": "https://api.github.com/repos/huggingface/transformers/labels/Help%20wanted", "name": "Help wanted", "color": "008672", "default": false, "description": "Extra attention is needed, help appreciated" }, { "id": 1260952223, "node_id": "MDU6TGFiZWwxMjYwOTUyMjIz", "url": "https://api.github.com/repos/huggingface/transformers/labels/Discussion", "name": "Discussion", "color": "22870e", "default": false, "description": "Discussion on a topic (keep it focused or open a new issue though)" }, { "id": 1314768611, "node_id": "MDU6TGFiZWwxMzE0NzY4NjEx", "url": "https://api.github.com/repos/huggingface/transformers/labels/wontfix", "name": "wontfix", "color": "ffffff", "default": true, "description": null } ]

[]

[ "I also encountered a similar problem AttributeError: 'BertForPreTraining' object has no attribute 'crf_loss'\r\n\r\n@thomwolf \r\n\r\nLooking forward to your reply", "Hi, from my reading of the [Natural Questions model](https://github.com/google-research/language/blob/master/language/question_answering/bert_joint/run_nq.py#L879-L932) it doesn't seems to be directly possible to load this model in the current library (with simple hacks).\r\n\r\nYou will need to define a new sub-class of `BertModel` (e.g. `BertForNaturalQA`) that reproduce the architecture of the TensorFlow model I pointed to. If you use the same name as the TensorFlow variables for the attributes of your PyTorch model you should be able to load the model with the current loading script.\r\n\r\nI don't have time to do this right now but if you want to start opening a PR, I can review it.\r\n\r\nBasically, just add another class after the `BertForQuestionAnswering` class in [`modeling.py`](https://github.com/huggingface/pytorch-pretrained-BERT/blob/master/pytorch_pretrained_bert/modeling.py#L1130)", "Thanks a lot for the reply. Will try this out.\r\n\r\nBest,\r\n", "This issue has been automatically marked as stale because it has not had recent activity. It will be closed if no further activity occurs. Thank you for your contributions.\n", "Hi @raheja, do you have any updates related to this? I'm trying to do the same and would welcome any hint! ", "Hi @paulachocron \r\nHaven't been able to try or implement this yet. I am still using the tf version for now. \r\n" ]

{ "url": "https://api.github.com/repos/huggingface/transformers/issues/449/reactions", "total_count": 1, "+1": 1, "-1": 0, "laugh": 0, "hooray": 0, "confused": 0, "heart": 0, "rocket": 0, "eyes": 0 }

{ "login": "Jason-kid", "id": 31425629, "node_id": "MDQ6VXNlcjMxNDI1NjI5", "avatar_url": "https://avatars.githubusercontent.com/u/31425629?v=4", "gravatar_id": "", "url": "https://api.github.com/users/Jason-kid", "html_url": "https://github.com/Jason-kid", "followers_url": "https://api.github.com/users/Jason-kid/followers", "following_url": "https://api.github.com/users/Jason-kid/following{/other_user}", "gists_url": "https://api.github.com/users/Jason-kid/gists{/gist_id}", "starred_url": "https://api.github.com/users/Jason-kid/starred{/owner}{/repo}", "subscriptions_url": "https://api.github.com/users/Jason-kid/subscriptions", "organizations_url": "https://api.github.com/users/Jason-kid/orgs", "repos_url": "https://api.github.com/users/Jason-kid/repos", "events_url": "https://api.github.com/users/Jason-kid/events{/privacy}", "received_events_url": "https://api.github.com/users/Jason-kid/received_events", "type": "User", "site_admin": false }

[]

[]

[ "Hi, you should probably turn to the TensorFlow version for pre-training.\r\nThis package is mostly intended to be used for fine-tuning pre-trained models.\r\nAnother option for BERT-like pre-training is to use Facebook's [XLM](https://github.com/facebookresearch/XLM)" ]

{ "url": "https://api.github.com/repos/huggingface/transformers/issues/448/reactions", "total_count": 2, "+1": 2, "-1": 0, "laugh": 0, "hooray": 0, "confused": 0, "heart": 0, "rocket": 0, "eyes": 0 }

{ "login": "zijwang", "id": 25057983, "node_id": "MDQ6VXNlcjI1MDU3OTgz", "avatar_url": "https://avatars.githubusercontent.com/u/25057983?v=4", "gravatar_id": "", "url": "https://api.github.com/users/zijwang", "html_url": "https://github.com/zijwang", "followers_url": "https://api.github.com/users/zijwang/followers", "following_url": "https://api.github.com/users/zijwang/following{/other_user}", "gists_url": "https://api.github.com/users/zijwang/gists{/gist_id}", "starred_url": "https://api.github.com/users/zijwang/starred{/owner}{/repo}", "subscriptions_url": "https://api.github.com/users/zijwang/subscriptions", "organizations_url": "https://api.github.com/users/zijwang/orgs", "repos_url": "https://api.github.com/users/zijwang/repos", "events_url": "https://api.github.com/users/zijwang/events{/privacy}", "received_events_url": "https://api.github.com/users/zijwang/received_events", "type": "User", "site_admin": false }

[]

[]

[ "Hi @zijwang , the package doesn't implement any specific batching logic, only tokenizers and models.\r\nYou are supposed to take care of this yourself in your scripts." ]

{ "url": "https://api.github.com/repos/huggingface/transformers/issues/447/reactions", "total_count": 0, "+1": 0, "-1": 0, "laugh": 0, "hooray": 0, "confused": 0, "heart": 0, "rocket": 0, "eyes": 0 }

{ "login": "yexing99", "id": 25463584, "node_id": "MDQ6VXNlcjI1NDYzNTg0", "avatar_url": "https://avatars.githubusercontent.com/u/25463584?v=4", "gravatar_id": "", "url": "https://api.github.com/users/yexing99", "html_url": "https://github.com/yexing99", "followers_url": "https://api.github.com/users/yexing99/followers", "following_url": "https://api.github.com/users/yexing99/following{/other_user}", "gists_url": "https://api.github.com/users/yexing99/gists{/gist_id}", "starred_url": "https://api.github.com/users/yexing99/starred{/owner}{/repo}", "subscriptions_url": "https://api.github.com/users/yexing99/subscriptions", "organizations_url": "https://api.github.com/users/yexing99/orgs", "repos_url": "https://api.github.com/users/yexing99/repos", "events_url": "https://api.github.com/users/yexing99/events{/privacy}", "received_events_url": "https://api.github.com/users/yexing99/received_events", "type": "User", "site_admin": false }

[ { "id": 1260952223, "node_id": "MDU6TGFiZWwxMjYwOTUyMjIz", "url": "https://api.github.com/repos/huggingface/transformers/labels/Discussion", "name": "Discussion", "color": "22870e", "default": false, "description": "Discussion on a topic (keep it focused or open a new issue though)" } ]

[]

[ "Hi @yexing99,\r\nWhat you need is the `hidden_state`, not the `weights`of the model.\r\nDon't use `model.named_parameters()` but just use the output of the model.\r\nHere is an example: https://github.com/huggingface/pytorch-pretrained-BERT#bert\r\nAnd more details here: https://github.com/huggingface/pytorch-pretrained-BERT#1-bertmodel\r\nYou should do somethings like this to get the BERT features from the second to the last hidden layer if the word `test` is for example the *fifth* word in your sentence (index 4 in the sentence):\r\n```python\r\nencoded_layers, _ = model(tokens_tensor, segments_tensors)\r\nfeatures = [encoded_layer[4] for encoded_layer in encoded_layers[1:]]\r\n```\r\n" ]

{ "url": "https://api.github.com/repos/huggingface/transformers/issues/446/reactions", "total_count": 0, "+1": 0, "-1": 0, "laugh": 0, "hooray": 0, "confused": 0, "heart": 0, "rocket": 0, "eyes": 0 }

{ "login": "lukovnikov", "id": 1732910, "node_id": "MDQ6VXNlcjE3MzI5MTA=", "avatar_url": "https://avatars.githubusercontent.com/u/1732910?v=4", "gravatar_id": "", "url": "https://api.github.com/users/lukovnikov", "html_url": "https://github.com/lukovnikov", "followers_url": "https://api.github.com/users/lukovnikov/followers", "following_url": "https://api.github.com/users/lukovnikov/following{/other_user}", "gists_url": "https://api.github.com/users/lukovnikov/gists{/gist_id}", "starred_url": "https://api.github.com/users/lukovnikov/starred{/owner}{/repo}", "subscriptions_url": "https://api.github.com/users/lukovnikov/subscriptions", "organizations_url": "https://api.github.com/users/lukovnikov/orgs", "repos_url": "https://api.github.com/users/lukovnikov/repos", "events_url": "https://api.github.com/users/lukovnikov/events{/privacy}", "received_events_url": "https://api.github.com/users/lukovnikov/received_events", "type": "User", "site_admin": false }

[]

[]

[ "Sorry for the delay in reviewing this.\r\nThis is a great PR and it looks good to me.\r\nThanks for adding some tests also.\r\nI agree with the (mostly cosmetic) comments from @marpaia.\r\nDo you think you can fix them and then we can merge?", "Fixed @marpaia 's comments.", "Awesome @lukovnikov, I think it looks great! Thanks for making this already awesome library even more awesome 💯 ", "Thanks @lukovnikov!" ]

{ "url": "https://api.github.com/repos/huggingface/transformers/issues/445/reactions", "total_count": 0, "+1": 0, "-1": 0, "laugh": 0, "hooray": 0, "confused": 0, "heart": 0, "rocket": 0, "eyes": 0 }

{ "url": "https://api.github.com/repos/huggingface/transformers/pulls/445", "html_url": "https://github.com/huggingface/transformers/pull/445", "diff_url": "https://github.com/huggingface/transformers/pull/445.diff", "patch_url": "https://github.com/huggingface/transformers/pull/445.patch", "merged_at": 1556008059000 }

{ "login": "alphanlp", "id": 12368732, "node_id": "MDQ6VXNlcjEyMzY4NzMy", "avatar_url": "https://avatars.githubusercontent.com/u/12368732?v=4", "gravatar_id": "", "url": "https://api.github.com/users/alphanlp", "html_url": "https://github.com/alphanlp", "followers_url": "https://api.github.com/users/alphanlp/followers", "following_url": "https://api.github.com/users/alphanlp/following{/other_user}", "gists_url": "https://api.github.com/users/alphanlp/gists{/gist_id}", "starred_url": "https://api.github.com/users/alphanlp/starred{/owner}{/repo}", "subscriptions_url": "https://api.github.com/users/alphanlp/subscriptions", "organizations_url": "https://api.github.com/users/alphanlp/orgs", "repos_url": "https://api.github.com/users/alphanlp/repos", "events_url": "https://api.github.com/users/alphanlp/events{/privacy}", "received_events_url": "https://api.github.com/users/alphanlp/received_events", "type": "User", "site_admin": false }

[ { "id": 1260952223, "node_id": "MDU6TGFiZWwxMjYwOTUyMjIz", "url": "https://api.github.com/repos/huggingface/transformers/labels/Discussion", "name": "Discussion", "color": "22870e", "default": false, "description": "Discussion on a topic (keep it focused or open a new issue though)" }, { "id": 1314768611, "node_id": "MDU6TGFiZWwxMzE0NzY4NjEx", "url": "https://api.github.com/repos/huggingface/transformers/labels/wontfix", "name": "wontfix", "color": "ffffff", "default": true, "description": null } ]

[]

[ "A CRF gives better NER F1 scores in some cases, but not necessarily in all cases. In the BERT paper, no CRF is used and hence also no CRF in this repository. I'd presume the BERT authors tested both with and without CRF and found that a CRF layer gives no improvement, since using a CRF is kind of the default setting nowadays.", "Issue #64 is a good reference for discussion on NER.", "This issue has been automatically marked as stale because it has not had recent activity. It will be closed if no further activity occurs. Thank you for your contributions.\n", "This repository have showed how to add a CRF layer on transformers to get a better performance on token classification task.\r\nhttps://github.com/shushanxingzhe/transformers_ner", "> github.com/shushanxingzhe/transformers_ner\r\n\r\nyour code does not work" ]

{ "url": "https://api.github.com/repos/huggingface/transformers/issues/444/reactions", "total_count": 0, "+1": 0, "-1": 0, "laugh": 0, "hooray": 0, "confused": 0, "heart": 0, "rocket": 0, "eyes": 0 }

{ "login": "shadylpstan", "id": 14815653, "node_id": "MDQ6VXNlcjE0ODE1NjUz", "avatar_url": "https://avatars.githubusercontent.com/u/14815653?v=4", "gravatar_id": "", "url": "https://api.github.com/users/shadylpstan", "html_url": "https://github.com/shadylpstan", "followers_url": "https://api.github.com/users/shadylpstan/followers", "following_url": "https://api.github.com/users/shadylpstan/following{/other_user}", "gists_url": "https://api.github.com/users/shadylpstan/gists{/gist_id}", "starred_url": "https://api.github.com/users/shadylpstan/starred{/owner}{/repo}", "subscriptions_url": "https://api.github.com/users/shadylpstan/subscriptions", "organizations_url": "https://api.github.com/users/shadylpstan/orgs", "repos_url": "https://api.github.com/users/shadylpstan/repos", "events_url": "https://api.github.com/users/shadylpstan/events{/privacy}", "received_events_url": "https://api.github.com/users/shadylpstan/received_events", "type": "User", "site_admin": false }

[ { "id": 1108649070, "node_id": "MDU6TGFiZWwxMTA4NjQ5MDcw", "url": "https://api.github.com/repos/huggingface/transformers/labels/Need%20more%20information", "name": "Need more information", "color": "d876e3", "default": false, "description": "Further information is requested" }, { "id": 1314768611, "node_id": "MDU6TGFiZWwxMzE0NzY4NjEx", "url": "https://api.github.com/repos/huggingface/transformers/labels/wontfix", "name": "wontfix", "color": "ffffff", "default": true, "description": null } ]

[]

[ "Maybe it's a problem of not setting `do_lower_case=False` and using a `cased` model like in #436.\r\nWhich pretrained model are you using?", "@thomwolf \r\nsorry but this is completely not what I am asking for.\r\nSuppose I have medical data and the names of the medicines and diseases are not in the bert pre-trained corpus. So bert gives me [UNK] token for important medical terms. How can I train it on my custom data?\r\n\r\nUsing lower cased multilingual pre-trained model.\r\n\r\n**PS: Medical data was just an example.** ", "This issue has been automatically marked as stale because it has not had recent activity. It will be closed if no further activity occurs. Thank you for your contributions.\n", "Hi, I'm also facing similar issue plz let me know if you have found any solution @shadylpstan \r\n", "> Hi, I'm also facing similar issue plz let me know if you have found any solution @shadylpstan\r\n\r\nI was successful in modifying a couple of bert files and train it using my corpus.\r\nThough it was really a mess doing so because by default bert has a classification model of a certain number of classes.\r\nThe accuracy wasn't good enough. So no idea it is a worth giving it shot from your end." ]

{ "url": "https://api.github.com/repos/huggingface/transformers/issues/443/reactions", "total_count": 1, "+1": 1, "-1": 0, "laugh": 0, "hooray": 0, "confused": 0, "heart": 0, "rocket": 0, "eyes": 0 }

{ "login": "AmrHendy", "id": 20549448, "node_id": "MDQ6VXNlcjIwNTQ5NDQ4", "avatar_url": "https://avatars.githubusercontent.com/u/20549448?v=4", "gravatar_id": "", "url": "https://api.github.com/users/AmrHendy", "html_url": "https://github.com/AmrHendy", "followers_url": "https://api.github.com/users/AmrHendy/followers", "following_url": "https://api.github.com/users/AmrHendy/following{/other_user}", "gists_url": "https://api.github.com/users/AmrHendy/gists{/gist_id}", "starred_url": "https://api.github.com/users/AmrHendy/starred{/owner}{/repo}", "subscriptions_url": "https://api.github.com/users/AmrHendy/subscriptions", "organizations_url": "https://api.github.com/users/AmrHendy/orgs", "repos_url": "https://api.github.com/users/AmrHendy/repos", "events_url": "https://api.github.com/users/AmrHendy/events{/privacy}", "received_events_url": "https://api.github.com/users/AmrHendy/received_events", "type": "User", "site_admin": false }

[ { "id": 1260952223, "node_id": "MDU6TGFiZWwxMjYwOTUyMjIz", "url": "https://api.github.com/repos/huggingface/transformers/labels/Discussion", "name": "Discussion", "color": "22870e", "default": false, "description": "Discussion on a topic (keep it focused or open a new issue though)" }, { "id": 1314768611, "node_id": "MDU6TGFiZWwxMzE0NzY4NjEx", "url": "https://api.github.com/repos/huggingface/transformers/labels/wontfix", "name": "wontfix", "color": "ffffff", "default": true, "description": null } ]

[]

[ "This issue has been automatically marked as stale because it has not had recent activity. It will be closed if no further activity occurs. Thank you for your contributions.\n" ]

{ "url": "https://api.github.com/repos/huggingface/transformers/issues/442/reactions", "total_count": 0, "+1": 0, "-1": 0, "laugh": 0, "hooray": 0, "confused": 0, "heart": 0, "rocket": 0, "eyes": 0 }

{ "login": "MottoX", "id": 6220861, "node_id": "MDQ6VXNlcjYyMjA4NjE=", "avatar_url": "https://avatars.githubusercontent.com/u/6220861?v=4", "gravatar_id": "", "url": "https://api.github.com/users/MottoX", "html_url": "https://github.com/MottoX", "followers_url": "https://api.github.com/users/MottoX/followers", "following_url": "https://api.github.com/users/MottoX/following{/other_user}", "gists_url": "https://api.github.com/users/MottoX/gists{/gist_id}", "starred_url": "https://api.github.com/users/MottoX/starred{/owner}{/repo}", "subscriptions_url": "https://api.github.com/users/MottoX/subscriptions", "organizations_url": "https://api.github.com/users/MottoX/orgs", "repos_url": "https://api.github.com/users/MottoX/repos", "events_url": "https://api.github.com/users/MottoX/events{/privacy}", "received_events_url": "https://api.github.com/users/MottoX/received_events", "type": "User", "site_admin": false }

[]

[]

[ "Hi @MottoX, sorry for the delay on reviewing this.\r\nIt seems to make sense to me.\r\nWhat kind of testing are you referring to? Could you share a bit more about them?", "Hi, @thomwolf \r\nI was trying to train a BERT-based model on NewsQA, a document-level QA dataset, using `run_squad.py`. I used 384 for max_seq_length and 128 for doc_stride. The model got an extremely bad result. I found there were numerous chunks with start_position and end_position as zeros.\r\nThen I checked the implementation of `run_squad.py` as well as `BertForQuestionAnswering`, which is basically BERT + one fully connected layer. I suspected that such out-of-span chunks are harmful for model to converge. So I modified the script to drop all these out-of-span chunks during training and got an acceptable result. I am sorry that I do not remember the exact numbers before and after doing this.\r\n\r\nI think the reason why the problem is not identified before is that context length (token-level) in SQuAD is not that long, which means the model can still easily converge anyway, with or without out-of-span chunks during training as they are really rare. However, when we have numerous out-of-span chunks in the training set, there will be a problem for model to converge.", "I looked at the code again, and found it is quite confusing, since chunks with `start_position==0` and `end_position==0` are also used for the cases where we have examples that do not have answer (for version_2_with_negative). So the current fix seems not compatible with this. Now I am not exactly sure whether we need to train the model with those out-of-span chunks. However, from my last experience, there should be a problem when we have too many such out-of-span chunks in the training set and the model is biased to predicting the result on the first token (CLS), despite the fact that `write_predictions` filters out those invalid predictions." ]

{ "url": "https://api.github.com/repos/huggingface/transformers/issues/441/reactions", "total_count": 0, "+1": 0, "-1": 0, "laugh": 0, "hooray": 0, "confused": 0, "heart": 0, "rocket": 0, "eyes": 0 }

{ "url": "https://api.github.com/repos/huggingface/transformers/pulls/441", "html_url": "https://github.com/huggingface/transformers/pull/441", "diff_url": "https://github.com/huggingface/transformers/pull/441.diff", "patch_url": "https://github.com/huggingface/transformers/pull/441.patch", "merged_at": null }

{ "login": "gkv91", "id": 8320832, "node_id": "MDQ6VXNlcjgzMjA4MzI=", "avatar_url": "https://avatars.githubusercontent.com/u/8320832?v=4", "gravatar_id": "", "url": "https://api.github.com/users/gkv91", "html_url": "https://github.com/gkv91", "followers_url": "https://api.github.com/users/gkv91/followers", "following_url": "https://api.github.com/users/gkv91/following{/other_user}", "gists_url": "https://api.github.com/users/gkv91/gists{/gist_id}", "starred_url": "https://api.github.com/users/gkv91/starred{/owner}{/repo}", "subscriptions_url": "https://api.github.com/users/gkv91/subscriptions", "organizations_url": "https://api.github.com/users/gkv91/orgs", "repos_url": "https://api.github.com/users/gkv91/repos", "events_url": "https://api.github.com/users/gkv91/events{/privacy}", "received_events_url": "https://api.github.com/users/gkv91/received_events", "type": "User", "site_admin": false }

[ { "id": 1260952223, "node_id": "MDU6TGFiZWwxMjYwOTUyMjIz", "url": "https://api.github.com/repos/huggingface/transformers/labels/Discussion", "name": "Discussion", "color": "22870e", "default": false, "description": "Discussion on a topic (keep it focused or open a new issue though)" }, { "id": 1314768611, "node_id": "MDU6TGFiZWwxMzE0NzY4NjEx", "url": "https://api.github.com/repos/huggingface/transformers/labels/wontfix", "name": "wontfix", "color": "ffffff", "default": true, "description": null } ]

[]

[ "I extract features like examples in extarct_features.py. But went I used these features(the last encoded_layers) as word embeddings, I got a worse result than using 300D Glove. I also used these features to compute the cos similarity for each word in sentences, I found that all values were around 0.6", "This issue has been automatically marked as stale because it has not had recent activity. It will be closed if no further activity occurs. Thank you for your contributions.\n", "Hi. I really need someone's help regarding using the bert model to extract the word embeddings: This script: https://github.com/huggingface/pytorch-pretrained-BERT/blob/master/examples/extract_features.py doesn't exist anymore.\r\n\r\nAny suggestions instead that replaced this script? \r\n" ]

{ "url": "https://api.github.com/repos/huggingface/transformers/issues/440/reactions", "total_count": 0, "+1": 0, "-1": 0, "laugh": 0, "hooray": 0, "confused": 0, "heart": 0, "rocket": 0, "eyes": 0 }

{ "login": "moinnadeem", "id": 813367, "node_id": "MDQ6VXNlcjgxMzM2Nw==", "avatar_url": "https://avatars.githubusercontent.com/u/813367?v=4", "gravatar_id": "", "url": "https://api.github.com/users/moinnadeem", "html_url": "https://github.com/moinnadeem", "followers_url": "https://api.github.com/users/moinnadeem/followers", "following_url": "https://api.github.com/users/moinnadeem/following{/other_user}", "gists_url": "https://api.github.com/users/moinnadeem/gists{/gist_id}", "starred_url": "https://api.github.com/users/moinnadeem/starred{/owner}{/repo}", "subscriptions_url": "https://api.github.com/users/moinnadeem/subscriptions", "organizations_url": "https://api.github.com/users/moinnadeem/orgs", "repos_url": "https://api.github.com/users/moinnadeem/repos", "events_url": "https://api.github.com/users/moinnadeem/events{/privacy}", "received_events_url": "https://api.github.com/users/moinnadeem/received_events", "type": "User", "site_admin": false }

[ { "id": 1108649070, "node_id": "MDU6TGFiZWwxMTA4NjQ5MDcw", "url": "https://api.github.com/repos/huggingface/transformers/labels/Need%20more%20information", "name": "Need more information", "color": "d876e3", "default": false, "description": "Further information is requested" }, { "id": 1314768611, "node_id": "MDU6TGFiZWwxMzE0NzY4NjEx", "url": "https://api.github.com/repos/huggingface/transformers/labels/wontfix", "name": "wontfix", "color": "ffffff", "default": true, "description": null } ]

[]

[ "Hi @moinnadeem,\r\nWhat is the hardware you are using and what is the exact command you are using to run DistributedDataParallel with the PyTorch launch module? ", "This issue has been automatically marked as stale because it has not had recent activity. It will be closed if no further activity occurs. Thank you for your contributions.\n" ]

{ "url": "https://api.github.com/repos/huggingface/transformers/issues/439/reactions", "total_count": 0, "+1": 0, "-1": 0, "laugh": 0, "hooray": 0, "confused": 0, "heart": 0, "rocket": 0, "eyes": 0 }

{ "login": "ghost", "id": 10137, "node_id": "MDQ6VXNlcjEwMTM3", "avatar_url": "https://avatars.githubusercontent.com/u/10137?v=4", "gravatar_id": "", "url": "https://api.github.com/users/ghost", "html_url": "https://github.com/ghost", "followers_url": "https://api.github.com/users/ghost/followers", "following_url": "https://api.github.com/users/ghost/following{/other_user}", "gists_url": "https://api.github.com/users/ghost/gists{/gist_id}", "starred_url": "https://api.github.com/users/ghost/starred{/owner}{/repo}", "subscriptions_url": "https://api.github.com/users/ghost/subscriptions", "organizations_url": "https://api.github.com/users/ghost/orgs", "repos_url": "https://api.github.com/users/ghost/repos", "events_url": "https://api.github.com/users/ghost/events{/privacy}", "received_events_url": "https://api.github.com/users/ghost/received_events", "type": "User", "site_admin": false }

[]

[]

[ "Hi @SandeepBhutani,\r\nCan you point me to the script you use for finetuning in Tensorflow?", "Hi @thomwolf : Thanks for reply. \r\nFine Tuning is done by mentioning do_train=True on run_squad.py (From google bert release github page: [https://github.com/google-research/bert](https://github.com/google-research/bert)) \r\nInternally, it calls `estimator.train(input_fn=train_input_fn, max_steps=num_train_steps)` \r\nFinetuning file was also same train-v1.1.json.. [https://rajpurkar.github.io/SQuAD-explorer/dataset/train-v1.1.json](https://rajpurkar.github.io/SQuAD-explorer/dataset/train-v1.1.json)\r\nSample header of train file is : \r\n`{\"data\": [{\"title\": \"University_of_Notre_Dame\", \"paragraphs\": [{\"context\": \"Architecturally` \r\n\r\n\r\nFollowing observation in case it is useful:\r\nWhile converting checkpoint of origional uncased bert_model.ckpt following log is printed:\r\n```\r\nLoading TF weight bert/encoder/layer_9/output/LayerNorm/gamma with shape [768]\r\nLoading TF weight bert/encoder/layer_9/output/dense/bias with shape [768]\r\nLoading TF weight bert/encoder/layer_9/output/dense/kernel with shape [3072, 768]\r\nLoading TF weight bert/pooler/dense/bias with shape [768]\r\nLoading TF weight bert/pooler/dense/kernel with shape [768, 768]\r\nLoading TF weight cls/predictions/output_bias with shape [30522]\r\nLoading TF weight cls/predictions/transform/LayerNorm/beta with shape [768]\r\nLoading TF weight cls/predictions/transform/LayerNorm/gamma with shape [768]\r\nLoading TF weight cls/predictions/transform/dense/bias with shape [768]\r\nLoading TF weight cls/predictions/transform/dense/kernel with shape [768, 768]\r\nLoading TF weight cls/seq_relationship/output_bias with shape [2]\r\nLoading TF weight cls/seq_relationship/output_weights with shape [2, 768]\r\nBuilding PyTorch model from configuration: {\r\n```\r\nWhile converting checkpoint after finetuning is done (model.ckpt-9000) following log is printed:\r\n```\r\nLoading TF weight bert/encoder/layer_9/output/dense/kernel/adam_m with shape [3072, 768]\r\nLoading TF weight bert/encoder/layer_9/output/dense/kernel/adam_v with shape [3072, 768]\r\nLoading TF weight bert/pooler/dense/bias with shape [768]\r\nLoading TF weight bert/pooler/dense/kernel with shape [768, 768]\r\nLoading TF weight cls/squad/output_bias with shape [2]\r\nLoading TF weight cls/squad/output_bias/adam_m with shape [2]\r\nLoading TF weight cls/squad/output_bias/adam_v with shape [2]\r\nLoading TF weight cls/squad/output_weights with shape [2, 768]\r\nLoading TF weight cls/squad/output_weights/adam_m with shape [2, 768]\r\nLoading TF weight cls/squad/output_weights/adam_v with shape [2, 768]\r\nLoading TF weight global_step with shape []\r\nBuilding PyTorch model from configuration: {\r\n```\r\n_cls/predictions_ is gone and _cls/squad_ appeared", "After reading the code of both tensorflow and pytorch version, figured out that tensorflow version is referring squad in create_model, like below (**cls/squad/output_weights**):\r\n```\r\ndef create_model(bert_config, is_training, input_ids, input_mask, segment_ids,\r\n use_one_hot_embeddings):\r\n \"\"\"Creates a classification model.\"\"\"\r\n model = modeling.BertModel(\r\n config=bert_config,\r\n is_training=is_training,\r\n input_ids=input_ids,\r\n input_mask=input_mask,\r\n token_type_ids=segment_ids,\r\n use_one_hot_embeddings=use_one_hot_embeddings)\r\n\r\n final_hidden = model.get_sequence_output()\r\n\r\n final_hidden_shape = modeling.get_shape_list(final_hidden, expected_rank=3)\r\n batch_size = final_hidden_shape[0]\r\n seq_length = final_hidden_shape[1]\r\n hidden_size = final_hidden_shape[2]\r\n\r\n output_weights = tf.get_variable(\r\n \"cls/squad/output_weights\", [2, hidden_size],\r\n initializer=tf.truncated_normal_initializer(stddev=0.02))\r\n\r\n output_bias = tf.get_variable(\r\n \"cls/squad/output_bias\", [2], initializer=tf.zeros_initializer())\r\n```\r\nAny suggestion, what should be tweaked? And where (create_model in tensorflow version should be changed or convert_tf_checkpoint_to_pytorch in pytorch version should be changed?) \r\n\r\nLooks like the definition of pytorch model (BertForPreTraining mentioned in conversion script) is different from tensorflow version, when fine tuned. That is why cls -> squad -> output_bias is not found. Is my understanding correct? If yes, is correct class already available which we can refer while conversion?", "Hi @thomwolf , \r\nTo make the conversion work, in modeling.py of pytorch version, I have added the class and 1 line of code in BertPreTrainingHeads below. After this conversion is happening. But I am not sure if I have done correct thing (_being a beginner in both tf and pytorch_). \r\nWould you like to validate/correct please. \r\n\r\n\r\n```\r\nclass SandeepSquadClass(nn.Module): ########this class sandeep added\r\n def __init__(self, config, bert_model_embedding_weights): \r\n super(SandeepSquadClass, self).__init__()\r\n self.weight = Variable(torch.ones(2, config.hidden_size), requires_grad=True) \r\n self.bias = Variable(torch.ones(2), requires_grad=True)\r\n \r\n def forward(self):\r\n print(\"What to do?\")\r\n \r\nclass BertPreTrainingHeads(nn.Module):\r\n def __init__(self, config, bert_model_embedding_weights):\r\n super(BertPreTrainingHeads, self).__init__()\r\n self.predictions = BertLMPredictionHead(config, bert_model_embedding_weights)\r\n #sandeep code below 3 apr\r\n self.squad = SandeepSquadClass(config, bert_model_embedding_weights) ###this line sandeep added\r\n self.seq_relationship = nn.Linear(config.hidden_size, 2)\r\n\r\n\r\n```", "Hi @SandeepBhutani, I pushed a commit to master which should help you do this kind of thing.\r\n\r\nFirst, switch to master by cloning the repo and then follow the following instructions:\r\n\r\nThe `convert_tf_checkpoint_to_pytorch` conversion script is made to create `BertForPretraining` model which is not your use case but you can load another type of model by reproducing the behavior of this script as follows:\r\n\r\n```python\r\nfrom pytorch_pretrained_bert import BertConfig, BertForTokenClassification, load_tf_weights_in_bert\r\n\r\n# Initialise a configuration according to your model\r\nconfig = BertConfig.from_pretrained('bert-XXX-XXX')\r\n\r\n# You will need to load a BertForTokenClassification model\r\nmodel = BertForTokenClassification(config)\r\n\r\n# Load weights from tf checkpoint\r\nload_tf_weights_in_bert(model, tf_checkpoint_path)\r\n\r\n# Save pytorch-model\r\nprint(\"Save PyTorch model to {}\".format(pytorch_dump_path))\r\ntorch.save(model.state_dict(), pytorch_dump_path)\r\n```", "Thanks @thomwolf ... After following change the checkpoint generated smoothly. \r\n```\r\n #model = BertForPreTraining(config) ##commented this\r\n model = BertForTokenClassification(config, 2) ## Added this\r\n```\r\nLet us give a try to run prediction using new .bin file. Hope the results would be same as using tensorflow version with .ckpt file. \r\nAppreciate 👍 ", "I downloaded tensorflow checkpoints for domain specific bert model and extracted the zip file into the folder pretrained_bert which contains the following the three files\r\n\r\n- model.ckpt.data-00000-of-00001\r\n- model.ckpt.index\r\n- model.ckpt.meta\r\n\r\nI used the following code to convert tensorflow checkpoints to pytorch\r\n\r\n```\r\nimport torch\r\n\r\nfrom pytorch_transformers.modeling_bert import BertConfig, BertForPreTraining, load_tf_weights_in_bert\r\n\r\n\r\ntf_checkpoint_path=\"pretrained_bert/model.ckpt\"\r\nbert_config_file = \"bert-base-cased-config.json\"\r\npytorch_dump_path=\"pytorch_bert\"\r\n\r\nconfig = BertConfig.from_json_file(bert_config_file)\r\nprint(\"Building PyTorch model from configuration: {}\".format(str(config)))\r\nmodel = BertForPreTraining(config)\r\n\r\n# Load weights from tf checkpoint\r\nload_tf_weights_in_bert(model, config, tf_checkpoint_path)\r\n\r\n# Save pytorch-model\r\nprint(\"Save PyTorch model to {}\".format(pytorch_dump_path))\r\ntorch.save(model.state_dict(), pytorch_dump_path)\r\n```\r\n\r\nI got this error when I ran the above code\r\n\r\n**NotFoundError: Unsuccessful TensorSliceReader constructor: Failed to find any matching files for pretrained_bert/model.ckpt**\r\n\r\nAny help is really appreciated............", "Seems like the script cannot find your checkpoint. Try giving it the full absolute path to the file.", "@thomwolf \r\nThanks, I didn't get any error when I gave absolute path of the file. ", "I was trying to convert my fine tuned model to pytorch using the following command.\r\n\r\n`\r\ntf_checkpoint_path='models/model.ckpt-21'\r\nbert_config_file='PRETRAINED_MODELS/uncased_L-12_H-768_A-12/bert_config.json'\r\npytorch_dump_path='pytorch_models/pytorch_model.bin'\r\n\r\npython convert_bert_original_tf_checkpoint_to_pytorch.py --tf_checkpoint_path=$tf_checkpoint_path --bert_config_file=$bert_config_file --pytorch_dump_path=$pytorch_dump_path `\r\n\r\nThe issue that I face is given below. Any help would be appreciated.\r\n\r\nTraceback (most recent call last):\r\nFile \"convert_bert_original_tf_checkpoint_to_pytorch.py\", line 65, in\r\nargs.pytorch_dump_path)\r\nFile \"convert_bert_original_tf_checkpoint_to_pytorch.py\", line 36, in convert_tf_checkpoint_to_pytorch\r\nload_tf_weights_in_bert(model, config, tf_checkpoint_path)\r\nFile \"/home/cibin/virtual_envs/pytorch/lib/python3.7/site-packages/transformers/modeling_bert.py\", line 98, in load_tf_weights_in_bert\r\npointer = getattr(pointer, 'classifier')\r\nFile \"/home/cibin/virtual_envs/pytorch/lib/python3.7/site-packages/torch/nn/modules/module.py\", line 585, in getattr\r\ntype(self).name, name))\r\nAttributeError: 'BertPreTrainingHeads' object has no attribute 'classifier'", "**A possible solution if you're copying a SQuAD-fine-tuned Bert from TF to PT**\r\n\r\nIssue: \r\n`AttributeError: 'BertPreTrainingHeads' object has no attribute 'classifier'`\r\n\r\n\r\nIt works for me by doing the following steps:\r\n\r\nStep 1. \r\nIn the script `convert_tf_checkpoint_to_pytorch.py` (or `convert_bert_original_tf_checkpoint_to_pytorch.py`):\r\n\r\n- Replace all `BertForPreTraining `with `BertForQuestionAnswering`.\r\n\r\nStep 2.\r\nOpen the source code file `modeling_bert.py` in your package `site-packages\\transformers`:\r\n\r\n- In the function `load_tf_weights_in_bert`, replace\r\n`elif l[0] == 'squad':`\r\n `pointer = getattr(pointer, 'classifier')`\r\nwith\r\n`elif l[0] == 'squad':`\r\n `pointer = getattr(pointer, 'qa_outputs')`\r\n\r\nIt should work since `qa_outputs` is the attribute name for the output layer of `BertForQuestionAnswering` instead of `classifier`.\r\n\r\nStep 3.\r\nAfter copying, check your pytorch model by evaluating the `dev-v2.0.json` with a script like this:\r\n`python run_squad.py --model_type bert --model_name_or_path MODEL_PATH --do_eval --train_file None --predict_file dev-v2.0.json --max_seq_length 384 --doc_stride 128 --output_dir ./output/ --version_2_with_negative`\r\nwhere `output_dir` should contain a copy of the pytorch model.\r\n\r\nThis will result in an evaluation like this:\r\n`{\r\n \"exact\": 72.99755748336563,\r\n \"f1\": 76.24686988414918,\r\n \"total\": 11873,\r\n \"HasAns_exact\": 72.82388663967612,\r\n \"HasAns_f1\": 79.33182964482165,\r\n \"HasAns_total\": 5928,\r\n \"NoAns_exact\": 73.17073170731707,\r\n \"NoAns_f1\": 73.17073170731707,\r\n \"NoAns_total\": 5945,\r\n \"best_exact\": 74.3619978101575,\r\n \"best_exact_thresh\": -3.6369030475616455,\r\n \"best_f1\": 77.12234803941384,\r\n \"best_f1_thresh\": -3.6369030475616455\r\n}`\r\nfor a `BERT-Base` model.\r\n\r\nHowever, if using `BertForTokenClassification` instead, the model will not be correctly copied since the structures for the classification layer are different. I tried this and got a model that had a f1 score of 10%.", "AttributeError: 'BertForTokenClassification' object has no attribute 'predict'\r\nHow do I use BERT trained model for prediction?", "@rashibudati, please take a look at the docs, namely the [Usage](https://huggingface.co./transformers/usage.html#named-entity-recognition) section which shows how to use token classification models.", "@Hya-cinthus Thank you so much! This saved me a lot of headache! " ]

{ "url": "https://api.github.com/repos/huggingface/transformers/issues/438/reactions", "total_count": 0, "+1": 0, "-1": 0, "laugh": 0, "hooray": 0, "confused": 0, "heart": 0, "rocket": 0, "eyes": 0 }

{ "login": "MottoX", "id": 6220861, "node_id": "MDQ6VXNlcjYyMjA4NjE=", "avatar_url": "https://avatars.githubusercontent.com/u/6220861?v=4", "gravatar_id": "", "url": "https://api.github.com/users/MottoX", "html_url": "https://github.com/MottoX", "followers_url": "https://api.github.com/users/MottoX/followers", "following_url": "https://api.github.com/users/MottoX/following{/other_user}", "gists_url": "https://api.github.com/users/MottoX/gists{/gist_id}", "starred_url": "https://api.github.com/users/MottoX/starred{/owner}{/repo}", "subscriptions_url": "https://api.github.com/users/MottoX/subscriptions", "organizations_url": "https://api.github.com/users/MottoX/orgs", "repos_url": "https://api.github.com/users/MottoX/repos", "events_url": "https://api.github.com/users/MottoX/events{/privacy}", "received_events_url": "https://api.github.com/users/MottoX/received_events", "type": "User", "site_admin": false }

[]

[]

[]

{ "url": "https://api.github.com/repos/huggingface/transformers/issues/437/reactions", "total_count": 0, "+1": 0, "-1": 0, "laugh": 0, "hooray": 0, "confused": 0, "heart": 0, "rocket": 0, "eyes": 0 }

{ "url": "https://api.github.com/repos/huggingface/transformers/pulls/437", "html_url": "https://github.com/huggingface/transformers/pull/437", "diff_url": "https://github.com/huggingface/transformers/pull/437.diff", "patch_url": "https://github.com/huggingface/transformers/pull/437.patch", "merged_at": 1554198047000 }

{ "login": "chiehminwei", "id": 2521639, "node_id": "MDQ6VXNlcjI1MjE2Mzk=", "avatar_url": "https://avatars.githubusercontent.com/u/2521639?v=4", "gravatar_id": "", "url": "https://api.github.com/users/chiehminwei", "html_url": "https://github.com/chiehminwei", "followers_url": "https://api.github.com/users/chiehminwei/followers", "following_url": "https://api.github.com/users/chiehminwei/following{/other_user}", "gists_url": "https://api.github.com/users/chiehminwei/gists{/gist_id}", "starred_url": "https://api.github.com/users/chiehminwei/starred{/owner}{/repo}", "subscriptions_url": "https://api.github.com/users/chiehminwei/subscriptions", "organizations_url": "https://api.github.com/users/chiehminwei/orgs", "repos_url": "https://api.github.com/users/chiehminwei/repos", "events_url": "https://api.github.com/users/chiehminwei/events{/privacy}", "received_events_url": "https://api.github.com/users/chiehminwei/received_events", "type": "User", "site_admin": false }

[]

[]

[ "Hi @icewing1996,\r\nThis comes from the fact that your tokenizer has `do_lower_case=True` but you load an uncased model.\r\n\r\nTry loading the tokenizer like this `tok = BertTokenizer.from_pretrained('bert-base-multilingual-uncased', do_lower_case=False)`.\r\n\r\nThis is actually a common issue and I see Jacob has added [a test](https://github.com/google-research/bert/blob/master/tokenization.py#L28) in the google repo to check the coherence between the lower casing option and the model checkpoint name. I will add a similar test.", "I used the Korean wiki corpus with the \"do_lower_case = True\" option to create pretrain data.\r\nMost of the data is generated as [UNK] token as followings:\r\n\r\nINFO:tensorflow:next_sentence_labels: 1\r\nINFO:tensorflow:*** Example ***\r\nINFO:tensorflow:tokens: [CLS] gr ##ace [UNK] [MASK] [UNK] [UNK] [UNK] 2 ᄀ ##ᅢ [UNK] [UNK] [UNK] [UNK] gr ##ace - 1 ᄀ ##ᅪ gr ##ace - 2 ᄀ ##ᅡ ( [UNK] e ##ss ##p - 2 a , e ##ss ##p - 2 [MASK] 유전적 [UNK] 220 km [UNK] ᄀ ##ᅥ ##ᄅ ##ᅵ [UNK] [UNK] ᄆ ##ᅧ [UNK] [UNK] [UNK] ᄆ ##ᅧ [UNK] , [UNK] , [UNK] [UNK] [UNK] ##댜 [MASK] [MASK] [UNK] [UNK] [UNK] [UNK] [UNK] [UNK] [UNK] [UNK] [UNK] . [UNK] [UNK] [UNK] , n ##as ##a , j ##pl , d ##l ##r [UNK] [UNK] [UNK] [MASK] [MASK] [UNK] [MASK] [MASK] [MASK] [MASK] [MASK] [UNK] j ##pl [UNK] [UNK] [UNK] [UNK] [UNK] [UNK] . 2002 [UNK] 3 [UNK] [UNK] [UNK] ᄀ ##ᅵ ##ᄌ ##ᅵ [UNK] [UNK] [UNK] [UNK] [UNK] 2017 [UNK] 10 [UNK] [MASK] ᄀ ##ᅡ [UNK] [UNK] [UNK] . [UNK] [UNK] [UNK] [UNK] [UNK] 5 [UNK] [UNK] [UNK] [UNK] [UNK] [UNK] [UNK] [UNK] 15 [UNK] [UNK] [UNK] [UNK] [UNK] [UNK] [UNK] ᄇ ##ᅩ ##ᄋ ##ᅵ ##ᄌ ##ᅥ [UNK] [UNK] [UNK] [UNK] [UNK] [UNK] [UNK] [UNK] [UNK] . [SEP] [UNK] [MASK] [UNK] 1956 [UNK] [UNK] [UNK] [MASK] [UNK] [UNK] [UNK] [UNK] [UNK] [MASK] ᄀ ##ᅪ [UNK] [UNK] [UNK] [UNK] [UNK] ᄆ ##ᅩ ##ᄃ ##ᅮ 11 [UNK] [UNK] [UNK] [UNK] [MASK] ##졌 [MASK] [MASK] ᄀ ##ᅩ [UNK] [UNK] . [UNK] [UNK] [UNK] 3 [UNK] · [UNK] [MASK] [MASK] ᄀ ##ᅲ ##ᄆ ##ᅩ ᄅ ##ᅩ [UNK] [UNK] [UNK] [UNK] ᄇ ##ᅩ [UNK] [UNK] [UNK] [UNK] ( 人 ) [UNK] [UNK] [UNK] [UNK] [UNK] [UNK] [UNK] [UNK] . ᄀ ##ᅳ [MASK] [MASK] ᄀ ##ᅪ [UNK] , [MASK] [MASK] [MASK] [UNK] , [UNK] , ##ᅢ [UNK] [UNK] 착취 [UNK] [UNK] [UNK] [UNK] [UNK] [UNK] [UNK] [UNK] [MASK] [UNK] . [UNK] ( 任 忠 [UNK] , ? - ? ) [UNK] [UNK] [UNK] [UNK] [UNK] [UNK] [UNK] [UNK] [MASK] [UNK] [UNK] , [UNK] 토지 [UNK] . [UNK] [MASK] [MASK] [MASK] [MASK] [MASK] [MASK] [MASK] [MASK] [MASK] [MASK] [UNK] [UNK] [UNK] [MASK] [MASK] [UNK] . [UNK] [MASK] [MASK] [UNK] [UNK] ) [UNK] [UNK] [UNK] [UNK] [UNK] [UNK] [UNK] [MASK] [MASK] [UNK] [UNK] [UNK] [UNK] [UNK] , [UNK] , [MASK] [MASK] [UNK] [UNK] [UNK] [UNK] ᄀ ##ᅩ , [UNK] [UNK] [UNK] [UNK] ( [UNK] 1 [MASK] ) [UNK] [UNK] [UNK] [MASK] [UNK] [UNK] [UNK] [UNK] [UNK] [MASK] [UNK] [UNK] . [UNK] [UNK] [UNK] [UNK] [UNK] [UNK] 크라이 경성 ##렘 [MASK] [UNK] [UNK] [UNK] [UNK] 12 [UNK] ᄇ ##ᅮ ##ᄐ ##ᅥ [UNK] [UNK] [UNK] ##ᅡ ##ᄐ ##ᅡ ##ᄂ ##ᅡ [UNK] [UNK] [UNK] [UNK] . [UNK] 1914 [UNK] [UNK] [UNK] ᄀ ##ᅪ [MASK] [MASK] 올리 [UNK] [UNK] [UNK] [UNK] [UNK] [UNK] [MASK] [MASK] [MASK] [MASK] [UNK] 1 [UNK] ᄇ ##ᅮ ##ᄐ ##ᅥ [UNK] [UNK] [UNK] [UNK] [UNK] [UNK] [UNK] [MASK] [UNK] . [MASK] [MASK] [UNK] ᄀ ##ᅩ ##ᄌ ##ᅵ ##ᄃ ##ᅩ [UNK] [UNK] [MASK] ᄇ ##ᅡ , [UNK] [UNK] [UNK] [UNK] ᄇ ##ᅩ ##ᄃ ##ᅡ [UNK] [UNK] [UNK] [UNK] ᄅ ##ᅩ ᄂ ##ᅡ ##ᄐ ##ᅡ ##ᄂ ##ᅡ [UNK] [UNK] [MASK] . [MASK] [UNK] [UNK] [UNK] [UNK] [UNK] [UNK] [UNK] [UNK] . [UNK] [UNK] [UNK] [UNK] [UNK] [UNK] [SEP]", "> I'm using the pre-trained multilingual tokenizer to tokenize some Korean texts, but it seems like this tokenizer is unable to recognize any Korean text at all.\r\n> \r\n> For example, when running on all the Universal Dependency Korean treebanks, this tokenizer fails to tokenize (it produces '[UNK]') the following characters. I know some of them are not Korean, but most of them are. More confusingly, when I check the `vocab.txt` file for the [BERT-Base, Multilingual Cased](https://storage.googleapis.com/bert_models/2018_11_23/multi_cased_L-12_H-768_A-12.zip) in the original repo (https://github.com/google-research/bert), it shows that Korean characters (I manually checked for '렬') are in the vocabulary.\r\n> \r\n> This makes me wonder whether there is any bug in the tokenizer provided in this PyTorch repo. I'm considering switching back to the tokenizer in the original code base. Will this create any compatibility issues?\r\n> \r\n> {'燄', '렬', '툇', '촌', '내', '윙', '쌈', '꿇', '톨', '葚', '솟', '힌', '래', '銎', '凊', '컸', '톤', '', '졍', '의', '슭', '옷', '챠', '囒', '뷘', '멤', '싱', '츠', '령', '겸', '댄', '휙', '醗', '돕', '톳', '던', '페', '띔', '짚', '락', '앙', '왜', '핼', '컫', '쿵', '왠', '기', '냇', '칼', '런', '험', '드', '궐', '얄', '찜', '렝', '빈', '후', '싫', '雎', '덕', '틀', '딴', '굶', '뜸', 'ㅑ', '', '손', '펴', '쏠', '튀', '화', '슘', '대', '왕', '잣', '딨', '듈', '뎀', '鑛', '칭', '젝', '뉴', '', '쑤', '작', '묘', '飮', '새', '쑥', '嘈', '캐', '笳', '랩', '', '히', '맵', '냈', '츰', '꿨', '딤', '밑', '낼', '訇', '뀌', '뻑', '혔', '겝', '얹', '랍', '', '옅', '징', '권', '욱', '탭', '', '흩', '벳', '밈', '굴', '연', '엑', '綉', '숯', '무', '셋', '벌', '텃', '클', '쫑', '덟', '쳬', '멎', '칵', '팅', '뜨', '땄', '㎡', '끽', '뿍', '담', '펭', '쉰', '있', '쏘', '탤', '濞', '치', '옵', '잃', '언', '빽', '짠', '넙', '깆', '콤', '러', '픽', '푸', '회', '낍', '스', '쩨', '진', '풋', '횡', '탑', '솝', '딕', '영', '볐', '씹', '뒷', '퀼', '렸', '별', '옇', '얘', '謔', 'Ↄ', '팎', '긋', '뽐', '현', '켠', '룻', '소', '몇', '쨌', '벨', '쩐', '붐', '트', '떴', '', '줌', '꽤', '밭', '쓴', '매', '쳤', '뱃', '콰', '먕', '봤', '', '베', '첩', '년', '짐', '뻤', '괭', '함', '鍝', '도', '폭', '꾸', '뜩', '괞', '닦', '몹', '', '팰', '콕', '떻', '킬', '강', '찡', '궂', '꿉', '빴', '즈', '데', '껑', '엘', '약', '천', '씻', '', '㈜', '롯', '털', '야', '혈', '펀', '', '', '뺐', '繽', '닙', '쪽', '행', '넉', '휩', '퀀', '쇠', '셑', '댑', '얽', '질', '뻬', '', '덥', '骯', '뀔', '룩', '팀', '愙', '괜', '몽', '냐', '씬', '서', '또', '썬', '절', '갱', 'ㅊ', '왼', '펠', '놈', '쯩', '름', '토', '받', '璣', '咧', '츄', '깃', '헬', '잇', '덜', '곡', '것', '겹', '지', '檣', '억', '셀', '픔', '륙', '돈', 'ㅜ', '믈', '멈', '빗', '을', '둘', '탱', '얇', '식', '', '틱', '똑', '줄', '궤', '잼', '착', '廝', '枸', '볏', '샴', '疙', '飈', '', '댁', '날', '킵', '동', '룸', '', '헐', '부', '웬', '램', '윗', '숍', '齲', '빼', '빡', '혐', '뮌', '너', '란', '땠', '개', '이', '끔', '麩', '표', '훈', '간', '바', '옆', '틸', '탄', '살', 'ʈ', '꽂', '벙', '뱅', '닭', '멀', '줍', '✌', '짊', '葯', '쨍', '쥔', '않', '술', '육', '햇', '텁', '뵈', '쿄', '', '웅', '막', '낮', '브', '펼', '색', '갰', '께', '음', '쐬', '숱', '빛', '텝', '젠', '준', '키', '왓', '팝', '순', '떳', '번', '팡', '핫', '팬', '畹', '깍', '슷', '겋', '위', '꺽', '껀', '낸', '숴', '잠', '휴', '訐', '貮', '티', '늪', '깔', '涮', '촐', '좀', '쭐', '턴', '같', '먼', '꾹', '쎌', '충', '땅', '희', '敉', '캠', '넋', '맑', '갇', '凈', '를', '쌌', '잿', '똥', '핵', '올', '윈', '접', '뮤', '귤', '갖', '촘', '웍', '속', '댕', '砒', '謇', '郫', '낄', '죽', '가', '깁', '뷸', '는', '쏜', '뉜', '릉', '쓰', '텼', '환', '皰', '띠', '즙', '자', '톱', '봄', '갠', '쉽', '항', '답', '넣', '옐', '머', '빔', '쬐', '듄', '썰', '로', '칩', '마', '뽂', '숀', '침', '찍', '캘', '노', '뼈', '', '놔', '특', '곁', '챙', '밥', '형', '팠', '섬', '焗', '닉', '관', '쉐', '잡', '슐', '뗏', '획', '추', '욘', '꾼', '좌', '쫀', '나', '킹', '뼛', '은', '랴', '깻', '눈', '욕', '하', '', '찢', '젊', '궁', '총', '놋', '둬', '뮬', '詝', '낭', '얌', '셨', '첫', '柢', '셉', '', '닥', '결', '겪', '', '⅛', '뾰', '뢰', '륨', '좇', '꼐', '옌', '둔', '', '금', '평', '면', '癤', '암', '허', '뱉', '銠', '꿔', '✔', '멘', '겟', '싯', '씁', '림', '볕', '넨', '휠', '☝', '仚', '쟈', '덤', '벅', '즘', '숙', '철', 'ㄹ', '재', '낚', '쉬', '롬', '휘', '렀', '荑', '득', '옮', '렙', '신', '랏', '헉', '푼', '났', '퐁', '쌥', '볶', '밝', '샀', '뽑', '춘', '앓', '곧', '찮', '믿', '늙', '녘', '쌤', '예', '밍', '요', '엣', '딱', '못', '뺨', '렉', '모', '쟁', '퍽', '鬪', '팜', 'ɟ', '틈', '체', '벚', '뛴', '훅', '윤', '짤', '샐', '궈', '댐', '룡', '덧', '깊', '賬', '택', '넬', '흰', '髖', '쁠', '투', '만', '창', '묽', '썼', '', '셩', '僊', '수', '豢', '癥', '삼', '춰', '렁', '퀴', '슛', '쌉', '', '묶', '정', '탓', '얏', '', '버', '쯔', '밉', '챈', '♩', '리', '떨', '료', '했', '잘', '뤄', '붕', '쩌', '탁', '혀', '繙', '껴', '박', '᾽', '뜀', '홉', '팔', '컨', '饃', '釩', '웰', '꼿', '薺', '縕', '깜', '케', '세', '鎰', '褂', '슌', '벵', '꼈', '카', '甍', '檨', '寗', '빵', '찔', '엊', '오', '통', '', '꽝', '니', '첨', '맺', '랑', '헙', '믄', '더', '쳐', '🏼', '옳', '쁨', '鈇', '취', '玕', '럽', '푹', '눠', '제', '디', '벡', '메', '걔', '訄', '滹', '眈', '☎', '물', '갓', '없', '', '송', '뭇', '멸', '빅', '늉', '난', '휼', '흙', '깥', '옥', '쫙', '례', '폰', '맏', '되', '훨', '헝', '짧', '쪼', '▶', '끓', '톡', '흥', '홈', '및', '콘', '빠', '등', '르', '깝', '밟', '펑', '끼', '잭', '와', '넷', '율', '찼', '酃', '균', '뿜', '악', '튈', '률', '', '걱', '과', '퀄', '튜', '끈', '늑', '렐', '쫄', '멋', '―', '탈', '憮', '잉', '醺', '粦', '맣', '큘', '曧', '쿼', '파', '둡', '탕', '抔', '솥', '척', '고', '', '숭', '람', '벗', '잤', 'ௌ', '옻', '叨', '듀', '늘', '염', '늦', '돌', '껏', '귄', '', '봉', '움', '뭄', '낯', '템', '멍', '랜', '', '헌', '샹', '압', '몰', '㎢', '큼', '윽', '', '퀵', '덩', '잊', '眯', '피', '학', '넴', '갬', '녹', '출', '꼴', '', '퍈', '쎄', '굳', '외', '붓', '犂', '슴', '늬', '글', '룬', '탠', '듣', '엿', '칸', '뿌', '갸', '냄', '깡', '괴', '든', '불', '툰', '甪', '箬', '빳', '한', '晳', '킷', '쭉', '둑', '꿍', '쭈', '맙', '鷄', '뜯', '폄', '酩', '퓨', '조', '롤', '프', '뻔', '교', '쿨', '펜', '닯', '', '크', '뀐', '紈', '집', '땡', '놓', '훔', '즌', '헨', '녀', '켤', '따', '', '포', '땐', '랙', '콩', '합', '농', '', '전', '엥', '텄', '', '긍', '', '뮐', '슝', '꽃', '븐', '', '岦', '빤', '찐', '☺', '둥', '랬', '뚤', '깽', '뺏', '목', '슨', '웨', '굵', '칠', '뜻', '려', '산', '끕', '커', '튿', '단', '쌀', '햄', '牾', '켰', '륜', '밤', '골', '익', '흉', '뷔', '랄', '뱀', '숨', '돔', '붙', '찧', '둠', '갔', '럿', '져', 'ʂ', '할', '존', '챌', '늠', '픈', '뷰', '댈', '갑', '컹', '쪄', '맜', '騏', '칡', '餮', '앎', '누', '굿', '幪', '멕', '백', '곤', '싸', '꼬', '헷', '펄', '럴', '饕', '뚫', '芘', 'ᆢ', '젯', '릴', '북', '', '룹', '핀', '옛', '병', '낟', '품', '짱', '겼', '쉴', '앳', '닳', '쾌', '힐', '경', '짓', '델', '酊', '턱', '끌', '횃', '호', '씀', '켯', '‒', '컬', '밋', '인', '향', '렷', '蚺', '몸', '❄', '놀', '탯', '쏭', '왈', '점', '팟', '', '써', '념', '뒬', '혹', '폐', '격', '앨', '본', '듭', '엷', '헴', '루', '콜', '묏', '', '큰', '샛', '았', '돗', '퓌', '장', '깎', '훼', '엾', '맞', '윌', '윷', '졔', '꿰', '쩍', '양', '킨', '', '줬', '왔', '볍', '엌', '샤', '￡', '삶', '힘', '꽁', '길', '급', '럭', '엮', '瘩', '땀', '구', '쏟', '몫', '쫌', '✓', '으', '딛', '룰', '놉', '력', '齬', '뒀', '플', '꿩', '확', '졌', 'ㅇ', '㏊', 'ɖ', '넝', '빌', '여', '믹', '삭', '뽕', '뽈', '楂', '킴', '었', '캄', '사', '웠', '청', '뀨', '셈', '말', '넜', '', '맬', '삐', '紜', '챔', '', '논', '쥬', '딧', '쁘', '윔', '들', '뛸', '춧', '좋', '풍', '퓰', '뙤', '흽', '튼', '뮈', '끗', '공', '짖', '볼', '웃', '블', '랫', '뚜', '굉', '묵', '꿀', '톰', '객', '떤', '낙', '찾', '코', '셸', '녔', '원', '넥', '뗀', '펫', '셧', '袥', '羱', '심', '즉', 'ㅎ', '앤', '뜬', '냥', '塍', '쿠', '캔', '숲', '핏', '뭐', '댓', '먹', '얻', '엠', '됨', '띄', '뗄', '봇', '듐', '칫', '두', '롱', '능', '측', '崞', '융', '켓', '딩', '효', '㎝', '거', '샘', '변', '립', '맹', '괄', '쿰', '팸', '판', '앞', 'ㄷ', '밖', '켜', '밌', '', '맸', '', '임', '습', 'Ⓣ', '팽', '뻥', '릭', '높', '성', '짭', '첼', '덫', '슬', '迢', '월', '견', '미', '값', '터', '촉', '근', '옴', '김', '끝', '㎞', '센', '쏴', '呔', '엇', '떡', '뿔', '넛', '승', '섰', '곳', '직', '텅', '륭', '법', '잖', '', '겠', '곽', '용', '죠', '닿', '뭍', '', '생', '액', 'ㅁ', '탬', '뉘', '흡', '썽', '쟝', '♬', '낫', '폼', '층', '곬', '', '칙', '홀', '얕', '낡', '퇴', '臿', '춤', '꿋', '알', '잎', '략', '맡', '', '엔', '뫼', '우', '편', '팥', '였', '蛰', '괘', '鵪', '틴', '뇨', '녕', 'ৌ', '규', '튤', '흘', '', '췄', '딪', '郪', '유', '썹', '럼', '워', '선', '그', '씩', '셜', '떠', '캥', '갚', '뭡', '찰', '겁', '샌', '', '좁', '뇌', '끊', '계', '렘', '협', '밸', '멜', '안', '배', '솜', '귐', '챗', '셔', '꼼', '앗', '친', '꼽', '삿', '시', '각', '잽', '뎠', '울', '필', '', '찬', '쭙', '띤', '솔', '타', '꽹', '축', '밧', '騁', '텨', '완', '봐', '종', '僴', '箚', '발', '딸', '뗐', '깅', '摑', '십', '껄', '큽', '', '욤', '앉', '섯', '놨', '낳', '춥', '굼', '鬢', '독', '첸', '젼', '때', '堈', '錛', 'ɳ', '됩', '테', '죕', '잴', '보', '녁', '뺀', '쯤', '업', '렴', '릿', '懊', '잔', '숫', '웹', '뭉', '컴', '싶', '훌', '흑', '툴', '귓', '닻', '최', '❤', '빚', '겅', '뒹', '벽', '주', '쨋', '읽', '넌', '닮', '싼', '톈', '겨', '쐰', '냅', '듬', '째', '혼', '빻', '꺾', '므', '뚝', '꼭', '琺', '긁', '앵', '웸', '삽', '쥘', '납', '걷', '짰', '넓', '쓸', '걀', '헹', '쾰', '맘', '돋', '콸', '뛰', '걸', '분', 'ㅐ', '챘', '설', '砝', '갯', '룽', '온', '켄', '섭', '쫓', '쌩', '틋', '', '열', '퍼', 'ㅅ', '뒤', '채', '맨', '牘', '샵', '헤', '봅', '곯', '잌', '처', '굽', '극', '촨', '렌', '뎅', '폈', '응', '됐', '비', '차', '랭', '닐', '玷', '증', '섞', '컵', '량', '멩', '瓠', '졸', '녜', '빙', '딘', '썩', '눌', '잰', '뤼', '흠', '梃', '겐', '쑹', '남', '坮', '쩡', '묻', '롭', '췌', '겜', '국', '씨', '씽', '즐', '땟', '많', '덴', '氚', '쇼', '韆', '맴', '깬', '횟', '꺼', '렵', '얼', '반', '낀', '감', '榫', '줘', '죄', '얗', '린', '슈', 'ㅍ', '석', '님', '겉', '방', '벤', '☀', '팍', '꽉', '튕', '뻗', '쩔', '힙', '갈', '역', '듯', '邗', '컥', '牴', '돼', '셍', '붉', '흄', '₴', '', '젖', '달', '황', '홋', '눅', '퀘', '躓', '낱', '찌', '군', '촬', '', '느', '뻘', '눔', '갤', '헛', '입', '혁', '', '엎', '젓', '컷', '몬', '록', '냑', '아', '', '嚨', '애', '릅', '짝', '裋', '짜', '큐', '씌', '덮', '얀', '뜰', '론', '쌍', '앱', '쩝', '눕', '팩', '찹', '혜', '며', '텔', '뮨', '네', '싣', '샷', '곱', '', '빨', '헥', '딜', '옹', '흔', '킥', '태', '뿐', '에', '꾀', '껍', '겔', '끄', '퉁', '렇', '㎍', '닌', '샅', '싹', '', '륵', '縈', '흐', '짙', '뀜', '명', '핸', '', '뻐', '패', '짬', '범', '류', '쿤', '‐', '', '뎌', '될', '뚱', '셰', '쁜', '벼', '밀', '꿈', 'ㅋ', '다', '상', '라', '엽', '鞦', '舀', '닫', '쫘', '운', 'ㄴ', '어', '츌', '긴', '젤', '맥', '캉', '풀', '댔', '읍', '氘', '련', '훗', '쌓', '해', '곰', '뭘', '망', '저', '깨', '건', '당', '뽀', '섹', '활', '샨', '랗', '밴', '릎', '캡', '嫘', '널', '핍', '嘮', '앴', '酆', '', '중', '詼', '책', '엉', '된', 'ㅓ', '켈', '툼', '까', '실', '젭', '랐', '봣', '일', '닷', '초', '콧', '쥐', '렛', '櫾', '핑', '漶', '썸', '광', '넘', '떼', '탐', '곶', '쩜', '적', '쇄', '폴', '맛', '링', '레', '복', '띈', '늄', '텍', '뭔', '닝', '愨', '콥', '꿎', '참', '족', '른', '훑', '읊', '엄', '', '섀', '게', '펙', '삯', '맷', '烺', '검', '귀', '텐', '껌', '멧', '잦', '갗', '민', '찻', '긔', '뽁', '겊', '갉', '', '션', '츨', '릇', '눴', '깐', '껭', '탔', '닛', '문', '홍', '섣', '낌', '냉', '쉼', '撘', '띨', '', '숟'}\r\n\r\nCould you resolve your problem? I have the same problem with 'bert-base-multilingual-cased' on Korean corpus." ]

{ "url": "https://api.github.com/repos/huggingface/transformers/issues/436/reactions", "total_count": 0, "+1": 0, "-1": 0, "laugh": 0, "hooray": 0, "confused": 0, "heart": 0, "rocket": 0, "eyes": 0 }

{ "login": "marpaia", "id": 927168, "node_id": "MDQ6VXNlcjkyNzE2OA==", "avatar_url": "https://avatars.githubusercontent.com/u/927168?v=4", "gravatar_id": "", "url": "https://api.github.com/users/marpaia", "html_url": "https://github.com/marpaia", "followers_url": "https://api.github.com/users/marpaia/followers", "following_url": "https://api.github.com/users/marpaia/following{/other_user}", "gists_url": "https://api.github.com/users/marpaia/gists{/gist_id}", "starred_url": "https://api.github.com/users/marpaia/starred{/owner}{/repo}", "subscriptions_url": "https://api.github.com/users/marpaia/subscriptions", "organizations_url": "https://api.github.com/users/marpaia/orgs", "repos_url": "https://api.github.com/users/marpaia/repos", "events_url": "https://api.github.com/users/marpaia/events{/privacy}", "received_events_url": "https://api.github.com/users/marpaia/received_events", "type": "User", "site_admin": false }

[]

[]

[ "Yes! Thanks @marpaia!" ]

{ "url": "https://api.github.com/repos/huggingface/transformers/issues/435/reactions", "total_count": 0, "+1": 0, "-1": 0, "laugh": 0, "hooray": 0, "confused": 0, "heart": 0, "rocket": 0, "eyes": 0 }

{ "url": "https://api.github.com/repos/huggingface/transformers/pulls/435", "html_url": "https://github.com/huggingface/transformers/pull/435", "diff_url": "https://github.com/huggingface/transformers/pull/435.diff", "patch_url": "https://github.com/huggingface/transformers/pull/435.patch", "merged_at": 1554194501000 }

{ "login": "nicholasbailey87", "id": 9570481, "node_id": "MDQ6VXNlcjk1NzA0ODE=", "avatar_url": "https://avatars.githubusercontent.com/u/9570481?v=4", "gravatar_id": "", "url": "https://api.github.com/users/nicholasbailey87", "html_url": "https://github.com/nicholasbailey87", "followers_url": "https://api.github.com/users/nicholasbailey87/followers", "following_url": "https://api.github.com/users/nicholasbailey87/following{/other_user}", "gists_url": "https://api.github.com/users/nicholasbailey87/gists{/gist_id}", "starred_url": "https://api.github.com/users/nicholasbailey87/starred{/owner}{/repo}", "subscriptions_url": "https://api.github.com/users/nicholasbailey87/subscriptions", "organizations_url": "https://api.github.com/users/nicholasbailey87/orgs", "repos_url": "https://api.github.com/users/nicholasbailey87/repos", "events_url": "https://api.github.com/users/nicholasbailey87/events{/privacy}", "received_events_url": "https://api.github.com/users/nicholasbailey87/received_events", "type": "User", "site_admin": false }

[ { "id": 1108649053, "node_id": "MDU6TGFiZWwxMTA4NjQ5MDUz", "url": "https://api.github.com/repos/huggingface/transformers/labels/Help%20wanted", "name": "Help wanted", "color": "008672", "default": false, "description": "Extra attention is needed, help appreciated" }, { "id": 1260952223, "node_id": "MDU6TGFiZWwxMjYwOTUyMjIz", "url": "https://api.github.com/repos/huggingface/transformers/labels/Discussion", "name": "Discussion", "color": "22870e", "default": false, "description": "Discussion on a topic (keep it focused or open a new issue though)" }, { "id": 1314768611, "node_id": "MDU6TGFiZWwxMzE0NzY4NjEx", "url": "https://api.github.com/repos/huggingface/transformers/labels/wontfix", "name": "wontfix", "color": "ffffff", "default": true, "description": null } ]

[]

[ "Try train_batch_size =1. Alternatively I propose to finetune with the tensorflow model using colabs TPUs as these have far more memory.", "This issue has been automatically marked as stale because it has not had recent activity. It will be closed if no further activity occurs. Thank you for your contributions.\n" ]

{ "url": "https://api.github.com/repos/huggingface/transformers/issues/434/reactions", "total_count": 0, "+1": 0, "-1": 0, "laugh": 0, "hooray": 0, "confused": 0, "heart": 0, "rocket": 0, "eyes": 0 }

{ "login": "SeekPoint", "id": 18051187, "node_id": "MDQ6VXNlcjE4MDUxMTg3", "avatar_url": "https://avatars.githubusercontent.com/u/18051187?v=4", "gravatar_id": "", "url": "https://api.github.com/users/SeekPoint", "html_url": "https://github.com/SeekPoint", "followers_url": "https://api.github.com/users/SeekPoint/followers", "following_url": "https://api.github.com/users/SeekPoint/following{/other_user}", "gists_url": "https://api.github.com/users/SeekPoint/gists{/gist_id}", "starred_url": "https://api.github.com/users/SeekPoint/starred{/owner}{/repo}", "subscriptions_url": "https://api.github.com/users/SeekPoint/subscriptions", "organizations_url": "https://api.github.com/users/SeekPoint/orgs", "repos_url": "https://api.github.com/users/SeekPoint/repos", "events_url": "https://api.github.com/users/SeekPoint/events{/privacy}", "received_events_url": "https://api.github.com/users/SeekPoint/received_events", "type": "User", "site_admin": false }

[]

[]

[ "```\r\npython3 examples/lm_finetuning/simple_lm_finetuning.py \r\n--train_corpus sample_text.txt \r\n--bert_model bert-base-uncased \r\n--do_lower_case \r\n--output_dir finetuned_lm/\r\n```\r\nIn addition, you can refer to #385 ", "Yes let's keep a single issue on this. Closing in favor of #385." ]

{ "url": "https://api.github.com/repos/huggingface/transformers/issues/433/reactions", "total_count": 0, "+1": 0, "-1": 0, "laugh": 0, "hooray": 0, "confused": 0, "heart": 0, "rocket": 0, "eyes": 0 }

{ "login": "varun-nathan", "id": 48044871, "node_id": "MDQ6VXNlcjQ4MDQ0ODcx", "avatar_url": "https://avatars.githubusercontent.com/u/48044871?v=4", "gravatar_id": "", "url": "https://api.github.com/users/varun-nathan", "html_url": "https://github.com/varun-nathan", "followers_url": "https://api.github.com/users/varun-nathan/followers", "following_url": "https://api.github.com/users/varun-nathan/following{/other_user}", "gists_url": "https://api.github.com/users/varun-nathan/gists{/gist_id}", "starred_url": "https://api.github.com/users/varun-nathan/starred{/owner}{/repo}", "subscriptions_url": "https://api.github.com/users/varun-nathan/subscriptions", "organizations_url": "https://api.github.com/users/varun-nathan/orgs", "repos_url": "https://api.github.com/users/varun-nathan/repos", "events_url": "https://api.github.com/users/varun-nathan/events{/privacy}", "received_events_url": "https://api.github.com/users/varun-nathan/received_events", "type": "User", "site_admin": false }

[ { "id": 1108649070, "node_id": "MDU6TGFiZWwxMTA4NjQ5MDcw", "url": "https://api.github.com/repos/huggingface/transformers/labels/Need%20more%20information", "name": "Need more information", "color": "d876e3", "default": false, "description": "Further information is requested" }, { "id": 1314768611, "node_id": "MDU6TGFiZWwxMzE0NzY4NjEx", "url": "https://api.github.com/repos/huggingface/transformers/labels/wontfix", "name": "wontfix", "color": "ffffff", "default": true, "description": null } ]

[]

[ "You probably forgot to deactivate the DropOut modules with `model.eval()`", "Oh ok, I get it. That's helpful. Thanks!", "I tried that and now I seem to be getting the same predictions for any input. In other words, the logits don't change with change in input. Not sure why this is happening?", "Maybe a bug in your code?\r\nIf you can share a simple self-contained example exhibiting the behavior we can have a look.", "That's unlikely as I tried the same code with other problems and got reasonable results. The problems on which I tried this code include MRPC task, sentiment prediction on IMDB dataset and intent detection on smalltalk data. The results are logical and reasonable.\r\nThe dataset on which I get this behaviour reported above has about 20 examples for each of the 4 intents viz. restaurant_search(0), booking_table(1), greet(2) and thanks(3). So, I was thinking if this could be due to the data having less number of examples. The sequence classifier class is basically bert model + a single hidden layer neural network with output layer as the number of labels. I feel that there's not enough data to train the last (classifier) layer. Here's my test data:\r\ntest_df = pd.DataFrame({'text': [\"hey\", \"indian hotels\"],\r\n 'label': [2, 0]})\r\nI print the logits for each of the 2 test samples which are as follows:\r\ntensor([[ 28.9354, 28.3292, 20.1560, -20.7804]])\r\ntensor([[ 28.9354, 28.3292, 20.1560, -20.7804]])\r\n\r\nThe logits tensor doesn't change for any input text.\r\n", "This issue has been automatically marked as stale because it has not had recent activity. It will be closed if no further activity occurs. Thank you for your contributions.\n", "Hi @varun-nathan, were you ever able to solve this? I'm having a very similar issue with ReformerForSequenceClassification.", "Hi @varun-nathan and @jstremme, were either of you able to find the issue? I'm experiencing the same thing with ElectraForSequenceClassification!", "> Hi @varun-nathan and @jstremme, were either of you able to find the issue? I'm experiencing the same thing with ElectraForSequenceClassification!\r\n\r\nHi @mollha, I managed to solve this but can't remember exactly what I did. Did you try `model.eval()`? Also, make sure you are training on more than just a few sample records and using a large enough model in terms of neurons per layer etc. ", "Thanks! I was already using model.eval(), but my dataset size was too small (around 1000). After increasing to 15000 I am getting much better results.", "Excellent!" ]

{ "url": "https://api.github.com/repos/huggingface/transformers/issues/432/reactions", "total_count": 1, "+1": 1, "-1": 0, "laugh": 0, "hooray": 0, "confused": 0, "heart": 0, "rocket": 0, "eyes": 0 }

{ "login": "Archelunch", "id": 10900176, "node_id": "MDQ6VXNlcjEwOTAwMTc2", "avatar_url": "https://avatars.githubusercontent.com/u/10900176?v=4", "gravatar_id": "", "url": "https://api.github.com/users/Archelunch", "html_url": "https://github.com/Archelunch", "followers_url": "https://api.github.com/users/Archelunch/followers", "following_url": "https://api.github.com/users/Archelunch/following{/other_user}", "gists_url": "https://api.github.com/users/Archelunch/gists{/gist_id}", "starred_url": "https://api.github.com/users/Archelunch/starred{/owner}{/repo}", "subscriptions_url": "https://api.github.com/users/Archelunch/subscriptions", "organizations_url": "https://api.github.com/users/Archelunch/orgs", "repos_url": "https://api.github.com/users/Archelunch/repos", "events_url": "https://api.github.com/users/Archelunch/events{/privacy}", "received_events_url": "https://api.github.com/users/Archelunch/received_events", "type": "User", "site_admin": false }

[ { "id": 1260952223, "node_id": "MDU6TGFiZWwxMjYwOTUyMjIz", "url": "https://api.github.com/repos/huggingface/transformers/labels/Discussion", "name": "Discussion", "color": "22870e", "default": false, "description": "Discussion on a topic (keep it focused or open a new issue though)" }, { "id": 1314768611, "node_id": "MDU6TGFiZWwxMzE0NzY4NjEx", "url": "https://api.github.com/repos/huggingface/transformers/labels/wontfix", "name": "wontfix", "color": "ffffff", "default": true, "description": null } ]

[]

[ "Hi,\r\nYou can tokenize your data with the Transformer-XL tokenizer and use it to train your model.", "I understand it. But I want to fine tune pretrained model", "What is the difference?\r\n\r\nCan you start by trying to do what you would do normally to fine-tune a pytorch model and then if you encounter an issue give the error messages back here so we can help?", "I found example in readme\r\nAnd my question - can I fine tune model using this script with my own data? \r\n` python run_openai_gpt.py \\\r\n --model_name openai-gpt \\\r\n --do_train \\\r\n --do_eval \\\r\n --train_dataset $ROC_STORIES_DIR/cloze_test_val__spring2016\\ -\\ cloze_test_ALL_val.csv \\\r\n --eval_dataset $ROC_STORIES_DIR/cloze_test_test__spring2016\\ -\\ cloze_test_ALL_test.csv \\\r\n --output_dir ../log \\\r\n --train_batch_size 16 \\ `\r\n\r\nExample for gpt, but I guess it is the same for TransformerXL", "Hi, you will likely need to adapt this example since Transformer-XL uses memory cells but there is no ready to use example for fine-tuning Transformer-XL in the repo unfortunately (and I don't plan to add one in the near future).\r\n\r\nIf you want to give it a try feel free to ask more specific questions here.\r\n\r\nI would advise to start by reading carefully the author's paper to be sure you have a good understanding of the model. You can also probably start from the authors' own PyTorch training script (which I simplified to make the evaluation script in the present repo).", "Is it possible to use the transformer with non-tokenized data @thomwolf? I know the Transformer is built for language modeling, but I would like to take advantage of the self-attention aspect of the Transformer's self-attention mechanism to model continuous data, and wonder how doable that sounds to you? ", "This issue has been automatically marked as stale because it has not had recent activity. It will be closed if no further activity occurs. Thank you for your contributions.\n" ]