Commit
·
2d1694c
1
Parent(s):
27be6d7
update model card README.md
Browse files
README.md
ADDED
|
@@ -0,0 +1,82 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
---
|
| 2 |
+
license: apache-2.0
|
| 3 |
+
tags:
|
| 4 |
+
- summarization
|
| 5 |
+
- mT5_multilingual_XLSum
|
| 6 |
+
- mt5
|
| 7 |
+
- abstractive summarization
|
| 8 |
+
- ar
|
| 9 |
+
- xlsum
|
| 10 |
+
- generated_from_trainer
|
| 11 |
+
datasets:
|
| 12 |
+
- xlsum
|
| 13 |
+
model-index:
|
| 14 |
+
- name: mt5-base-finetune-ar-xlsum
|
| 15 |
+
results: []
|
| 16 |
+
---
|
| 17 |
+
|
| 18 |
+
<!-- This model card has been generated automatically according to the information the Trainer had access to. You
|
| 19 |
+
should probably proofread and complete it, then remove this comment. -->
|
| 20 |
+
|
| 21 |
+
# mt5-base-finetune-ar-xlsum
|
| 22 |
+
|
| 23 |
+
This model is a fine-tuned version of [google/mt5-base](https://huggingface.co/google/mt5-base) on the xlsum dataset.
|
| 24 |
+
It achieves the following results on the evaluation set:
|
| 25 |
+
- Loss: 3.2546
|
| 26 |
+
- Rouge-1: 22.2
|
| 27 |
+
- Rouge-2: 9.57
|
| 28 |
+
- Rouge-l: 20.26
|
| 29 |
+
- Gen Len: 19.0
|
| 30 |
+
- Bertscore: 71.43
|
| 31 |
+
|
| 32 |
+
## Model description
|
| 33 |
+
|
| 34 |
+
More information needed
|
| 35 |
+
|
| 36 |
+
## Intended uses & limitations
|
| 37 |
+
|
| 38 |
+
More information needed
|
| 39 |
+
|
| 40 |
+
## Training and evaluation data
|
| 41 |
+
|
| 42 |
+
More information needed
|
| 43 |
+
|
| 44 |
+
## Training procedure
|
| 45 |
+
|
| 46 |
+
### Training hyperparameters
|
| 47 |
+
|
| 48 |
+
The following hyperparameters were used during training:
|
| 49 |
+
- learning_rate: 0.0005
|
| 50 |
+
- train_batch_size: 4
|
| 51 |
+
- eval_batch_size: 4
|
| 52 |
+
- seed: 42
|
| 53 |
+
- gradient_accumulation_steps: 16
|
| 54 |
+
- total_train_batch_size: 64
|
| 55 |
+
- optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
|
| 56 |
+
- lr_scheduler_type: linear
|
| 57 |
+
- lr_scheduler_warmup_steps: 250
|
| 58 |
+
- num_epochs: 10
|
| 59 |
+
- label_smoothing_factor: 0.1
|
| 60 |
+
|
| 61 |
+
### Training results
|
| 62 |
+
|
| 63 |
+
| Training Loss | Epoch | Step | Validation Loss | Rouge-1 | Rouge-2 | Rouge-l | Gen Len | Bertscore |
|
| 64 |
+
|:-------------:|:-----:|:----:|:---------------:|:-------:|:-------:|:-------:|:-------:|:---------:|
|
| 65 |
+
| 4.9261 | 1.0 | 585 | 3.6314 | 18.19 | 6.49 | 16.37 | 19.0 | 70.17 |
|
| 66 |
+
| 3.8429 | 2.0 | 1170 | 3.4253 | 19.45 | 7.58 | 17.73 | 19.0 | 70.35 |
|
| 67 |
+
| 3.6311 | 3.0 | 1755 | 3.3569 | 20.83 | 8.54 | 18.9 | 19.0 | 70.89 |
|
| 68 |
+
| 3.4917 | 4.0 | 2340 | 3.3101 | 20.77 | 8.53 | 18.89 | 19.0 | 70.98 |
|
| 69 |
+
| 3.3873 | 5.0 | 2925 | 3.2867 | 21.47 | 9.0 | 19.54 | 19.0 | 71.23 |
|
| 70 |
+
| 3.3037 | 6.0 | 3510 | 3.2693 | 21.41 | 9.0 | 19.5 | 19.0 | 71.21 |
|
| 71 |
+
| 3.2357 | 7.0 | 4095 | 3.2581 | 22.05 | 9.36 | 20.04 | 19.0 | 71.43 |
|
| 72 |
+
| 3.1798 | 8.0 | 4680 | 3.2522 | 22.21 | 9.56 | 20.23 | 19.0 | 71.41 |
|
| 73 |
+
| 3.1359 | 9.0 | 5265 | 3.2546 | 22.27 | 9.58 | 20.23 | 19.0 | 71.46 |
|
| 74 |
+
| 3.0997 | 10.0 | 5850 | 3.2546 | 22.2 | 9.57 | 20.26 | 19.0 | 71.43 |
|
| 75 |
+
|
| 76 |
+
|
| 77 |
+
### Framework versions
|
| 78 |
+
|
| 79 |
+
- Transformers 4.19.4
|
| 80 |
+
- Pytorch 1.11.0+cu113
|
| 81 |
+
- Datasets 2.2.2
|
| 82 |
+
- Tokenizers 0.12.1
|