Fine tuning with synthetic data and translated. Batch Size = 32, 20 epochs, lr 5e3

README.md

@@ -0,0 +1,84 @@
+---
+library_name: transformers
+license: apache-2.0
+base_model: google/mt5-small
+tags:
+- summarization
+- generated_from_trainer
+metrics:
+- rouge
+model-index:
+- name: mt5-small-synthetic-data-plus-translated-bs32ep20lr5e3
+  results: []
+---
+
+<!-- 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. -->
+
+# mt5-small-synthetic-data-plus-translated-bs32ep20lr5e3
+
+This model is a fine-tuned version of [google/mt5-small](https://huggingface.co/google/mt5-small) on the None dataset.
+It achieves the following results on the evaluation set:
+- Loss: 1.3381
+- Rouge1: 0.7165
+- Rouge2: 0.6111
+- Rougel: 0.7004
+- Rougelsum: 0.7016
+
+## 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: 0.0056
+- train_batch_size: 32
+- eval_batch_size: 32
+- seed: 42
+- optimizer: Use OptimizerNames.ADAMW_TORCH with betas=(0.9,0.999) and epsilon=1e-08 and optimizer_args=No additional optimizer arguments
+- lr_scheduler_type: linear
+- num_epochs: 20
+
+### Training results
+
+| Training Loss | Epoch | Step | Validation Loss | Rouge1 | Rouge2 | Rougel | Rougelsum |
+|:-------------:|:-----:|:----:|:---------------:|:------:|:------:|:------:|:---------:|
+| 0.2243        | 1.0   | 38   | 0.9621          | 0.6801 | 0.5630 | 0.6599 | 0.6606    |
+| 0.2209        | 2.0   | 76   | 0.9423          | 0.6766 | 0.5707 | 0.6633 | 0.6644    |
+| 0.1953        | 3.0   | 114  | 0.9503          | 0.6525 | 0.5271 | 0.6361 | 0.6369    |
+| 0.1812        | 4.0   | 152  | 0.9818          | 0.6811 | 0.5742 | 0.6672 | 0.6680    |
+| 0.1418        | 5.0   | 190  | 0.9591          | 0.6868 | 0.5781 | 0.6700 | 0.6708    |
+| 0.1312        | 6.0   | 228  | 1.0121          | 0.6900 | 0.5842 | 0.6734 | 0.6742    |
+| 0.1236        | 7.0   | 266  | 0.9913          | 0.6787 | 0.5689 | 0.6652 | 0.6653    |
+| 0.1068        | 8.0   | 304  | 0.9773          | 0.6886 | 0.5781 | 0.6749 | 0.6764    |
+| 0.106         | 9.0   | 342  | 1.0201          | 0.6947 | 0.5825 | 0.6798 | 0.6802    |
+| 0.084         | 10.0  | 380  | 1.0865          | 0.6861 | 0.5775 | 0.6726 | 0.6738    |
+| 0.0744        | 11.0  | 418  | 1.0310          | 0.6997 | 0.5865 | 0.6849 | 0.6861    |
+| 0.0618        | 12.0  | 456  | 1.1647          | 0.7118 | 0.6182 | 0.7016 | 0.7020    |
+| 0.0493        | 13.0  | 494  | 1.1808          | 0.7089 | 0.6098 | 0.6959 | 0.6970    |
+| 0.0472        | 14.0  | 532  | 1.2040          | 0.7087 | 0.6090 | 0.6956 | 0.6965    |
+| 0.0399        | 15.0  | 570  | 1.1293          | 0.7065 | 0.6035 | 0.6953 | 0.6965    |
+| 0.0346        | 16.0  | 608  | 1.2286          | 0.7078 | 0.6028 | 0.6928 | 0.6940    |
+| 0.0255        | 17.0  | 646  | 1.2970          | 0.7114 | 0.6069 | 0.6986 | 0.7001    |
+| 0.0241        | 18.0  | 684  | 1.3016          | 0.7053 | 0.5983 | 0.6893 | 0.6904    |
+| 0.0217        | 19.0  | 722  | 1.3315          | 0.7137 | 0.6084 | 0.6999 | 0.7008    |
+| 0.0196        | 20.0  | 760  | 1.3381          | 0.7165 | 0.6111 | 0.7004 | 0.7016    |
+
+
+### Framework versions
+
+- Transformers 4.47.1
+- Pytorch 2.5.1+cu121
+- Datasets 3.2.0
+- Tokenizers 0.21.0

generation_config.json

@@ -0,0 +1,6 @@
+{
+  "decoder_start_token_id": 0,
+  "eos_token_id": 1,
+  "pad_token_id": 0,
+  "transformers_version": "4.47.1"
+}

runs/Jan27_18-22-48_83395d5f1f6d/events.out.tfevents.1738002490.83395d5f1f6d.12721.4

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:27bf6e4b940091f022ffeea296e1215faf7b98caa903375f48af7049d8d4d47a
+size 562