Add fine-tuned checkpoint

README.md

@@ -0,0 +1,73 @@
+---
+tags:
+- generated_from_trainer
+datasets:
+- Graphcore/vqa-lxmert
+metrics:
+- accuracy
+model-index:
+- name: vqa
+  results:
+  - task:
+      name: Question Answering
+      type: question-answering
+    dataset:
+      name: Graphcore/vqa-lxmert
+      type: Graphcore/vqa-lxmert
+      args: vqa
+    metrics:
+    - name: Accuracy
+      type: accuracy
+      value: 0.7242196202278137
+---
+
+<!-- This model card has been generated automatically according to the information the Trainer had access to. You
+should probably proofread and complete it, then remove this comment. -->
+
+# vqa
+
+This model is a fine-tuned version of [unc-nlp/lxmert-base-uncased](https://huggingface.co/unc-nlp/lxmert-base-uncased) on the Graphcore/vqa-lxmert dataset.
+It achieves the following results on the evaluation set:
+- Loss: 0.0009
+- Accuracy: 0.7242
+
+## Model description
+
+More information needed
+
+## Intended uses & limitations
+
+More information needed
+
+## Training and evaluation data
+
+More information needed
+
+## Training procedure
+
+### Training hyperparameters
+
+The following hyperparameters were used during training:
+- learning_rate: 5e-05
+- train_batch_size: 1
+- eval_batch_size: 8
+- seed: 42
+- distributed_type: IPU
+- total_train_batch_size: 64
+- total_eval_batch_size: 128
+- optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
+- lr_scheduler_type: linear
+- lr_scheduler_warmup_ratio: 0.1
+- num_epochs: 4.0
+- training precision: Mixed Precision
+
+### Training results
+
+
+
+### Framework versions
+
+- Transformers 4.18.0.dev0
+- Pytorch 1.10.0+cpu
+- Datasets 2.0.0
+- Tokenizers 0.11.6

all_results.json

@@ -0,0 +1,14 @@
+{
+    "epoch": 4.0,
+    "eval_accuracy": 0.7242196202278137,
+    "eval_loss": 0.0008745193481445312,
+    "eval_runtime": 1148.6856,
+    "eval_samples": 214354,
+    "eval_samples_per_second": 186.608,
+    "eval_steps_per_second": 1.458,
+    "train_loss": 0.0060005393999575125,
+    "train_runtime": 13854.802,
+    "train_samples": 443757,
+    "train_samples_per_second": 128.116,
+    "train_steps_per_second": 2.002
+}

config.json

@@ -0,0 +1,41 @@
+{
+  "_name_or_path": "unc-nlp/lxmert-base-uncased",
+  "architectures": [
+    "PoptorchPipelinedLxmertForQuestionAnswering"
+  ],
+  "attention_probs_dropout_prob": 0.1,
+  "hidden_act": "gelu",
+  "hidden_dropout_prob": 0.1,
+  "hidden_size": 768,
+  "initializer_range": 0.02,
+  "intermediate_size": 3072,
+  "l_layers": 9,
+  "layer_norm_eps": 1e-12,
+  "max_position_embeddings": 512,
+  "model_type": "lxmert",
+  "num_attention_heads": 12,
+  "num_attr_labels": 400,
+  "num_hidden_layers": {
+    "cross_encoder": 5,
+    "language": 9,
+    "vision": 5
+  },
+  "num_object_labels": 1600,
+  "num_qa_labels": 3129,
+  "r_layers": 5,
+  "task_mask_lm": true,
+  "task_matched": true,
+  "task_obj_predict": true,
+  "task_qa": true,
+  "torch_dtype": "float16",
+  "transformers_version": "4.18.0.dev0",
+  "type_vocab_size": 2,
+  "visual_attr_loss": true,
+  "visual_feat_dim": 2048,
+  "visual_feat_loss": true,
+  "visual_loss_normalizer": 6.67,
+  "visual_obj_loss": true,
+  "visual_pos_dim": 4,
+  "vocab_size": 30522,
+  "x_layers": 5
+}

eval_results.json

@@ -0,0 +1,9 @@
+{
+    "epoch": 4.0,
+    "eval_accuracy": 0.7242196202278137,
+    "eval_loss": 0.0008745193481445312,
+    "eval_runtime": 1148.6856,
+    "eval_samples": 214354,
+    "eval_samples_per_second": 186.608,
+    "eval_steps_per_second": 1.458
+}

ipu_config.json

@@ -0,0 +1,26 @@
+{
+  "device_iterations": 2,
+  "embedding_serialization_factor": 0,
+  "enable_half_first_order_momentum": true,
+  "enable_half_partials": true,
+  "executable_cache_dir": "./exe_cache",
+  "gradient_accumulation_steps": 8,
+  "inference_device_iterations": 4,
+  "inference_replication_factor": 4,
+  "ipus_per_replica": 4,
+  "layers_per_ipu": [
+    1
+  ],
+  "matmul_proportion": [
+    0.25
+  ],
+  "optimizer_state_offchip": true,
+  "optimum_version": "1.0.0",
+  "output_mode": "final",
+  "profile_dir": "",
+  "recompute_checkpoint_every_layer": true,
+  "replicated_tensor_sharding": true,
+  "replication_factor": 4,
+  "seed": 42,
+  "use_popdist": false
+}

pytorch_model.bin

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:3f7c262e9a7e3a0086cbaa7a4ccd62cc7809bd72cd9b8ae8662a477f72d4e65c
+size 428008947

special_tokens_map.json

@@ -0,0 +1 @@
+{"unk_token": "[UNK]", "sep_token": "[SEP]", "pad_token": "[PAD]", "cls_token": "[CLS]", "mask_token": "[MASK]"}

tokenizer_config.json

@@ -0,0 +1 @@
+{"do_lower_case": true, "unk_token": "[UNK]", "sep_token": "[SEP]", "pad_token": "[PAD]", "cls_token": "[CLS]", "mask_token": "[MASK]", "tokenize_chinese_chars": true, "strip_accents": null, "model_max_length": 512, "special_tokens_map_file": "/home/eltoto/unc-nlp/special_tokens_map.json", "full_tokenizer_file": null, "name_or_path": "unc-nlp/lxmert-base-uncased", "tokenizer_class": "LxmertTokenizer"}

train_results.json

@@ -0,0 +1,8 @@
+{
+    "epoch": 4.0,
+    "train_loss": 0.0060005393999575125,
+    "train_runtime": 13854.802,
+    "train_samples": 443757,
+    "train_samples_per_second": 128.116,
+    "train_steps_per_second": 2.002
+}

training_args.bin

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:c0496ca4ecb25793c825290b9c33c3c4e81e1681b878b0dc49a5b183ca816b4c
+size 2671