initial commit

README.md

@@ -1,3 +1,91 @@
----
-license: cc
----
+---
+license: apache-2.0
+tags:
+- generated_from_trainer
+datasets:
+- mlsum
+metrics:
+- rouge
+model-index:
+- name: eval-mt5-base-aggressive
+  results:
+  - task:
+      name: Summarization
+      type: summarization
+    dataset:
+      name: mlsum tu
+      type: mlsum
+      args: tu
+    metrics:
+    - name: Rouge1
+      type: rouge
+      value: 47.4222
+---
+
+<!-- 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. -->
+
+# eval-mt5-base-aggressive
+
+This model is a fine-tuned version of [google/mt5-base](https://huggingface.co/google/mt5-base) on the mlsum tu dataset.
+It achieves the following results on the evaluation set:
+- Loss: 2.7801
+- Rouge1: 47.4222
+- Rouge2: 34.8624
+- Rougel: 42.2487
+- Rougelsum: 43.9494
+- Gen Len: 51.3525
+
+## 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: 0.0005
+- train_batch_size: 2
+- eval_batch_size: 4
+- seed: 42
+- distributed_type: multi-GPU
+- num_devices: 8
+- gradient_accumulation_steps: 4
+- total_train_batch_size: 64
+- total_eval_batch_size: 32
+- optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
+- lr_scheduler_type: linear
+- num_epochs: 10.0
+- label_smoothing_factor: 0.1
+
+### Training results
+
+| Training Loss | Epoch | Step  | Validation Loss | Rouge1  | Rouge2  | Rougel  | Rougelsum | Gen Len |
+|:-------------:|:-----:|:-----:|:---------------:|:-------:|:-------:|:-------:|:---------:|:-------:|
+| 3.084         | 1.0   | 3895  | 2.9282          | 31.6872 | 22.1113 | 29.2851 | 29.7608   | 18.9861 |
+| 2.9162        | 2.0   | 7790  | 2.8552          | 32.1716 | 22.5001 | 29.6845 | 30.1887   | 18.9938 |
+| 2.8149        | 3.0   | 11685 | 2.8089          | 32.5681 | 22.689  | 30.0409 | 30.5507   | 18.9959 |
+| 2.7325        | 4.0   | 15580 | 2.7948          | 33.1236 | 23.1775 | 30.5156 | 31.0461   | 18.9958 |
+| 2.6679        | 5.0   | 19475 | 2.7810          | 33.1766 | 23.162  | 30.4802 | 31.0527   | 18.9967 |
+| 2.6237        | 6.0   | 23370 | 2.7790          | 33.1118 | 23.2043 | 30.5064 | 31.0096   | 18.9978 |
+| 2.5711        | 7.0   | 27265 | 2.7801          | 33.2033 | 23.2957 | 30.59   | 31.1504   | 18.9979 |
+| 2.538         | 8.0   | 31160 | 2.7777          | 33.0256 | 23.0621 | 30.3818 | 30.978    | 18.998  |
+| 2.5           | 9.0   | 35055 | 2.7839          | 33.2288 | 23.2361 | 30.5421 | 31.1573   | 18.998  |
+| 2.4719        | 10.0  | 38950 | 2.7832          | 33.2098 | 23.2274 | 30.5164 | 31.1094   | 18.9981 |
+
+
+### Framework versions
+
+- Transformers 4.11.3
+- Pytorch 1.8.2+cu111
+- Datasets 1.14.0
+- Tokenizers 0.10.3

config.json

@@ -0,0 +1,29 @@
+{
+  "_name_or_path": "google/mt5-base",
+  "architectures": [
+    "MT5ForConditionalGeneration"
+  ],
+  "d_ff": 2048,
+  "d_kv": 64,
+  "d_model": 768,
+  "decoder_start_token_id": 0,
+  "dropout_rate": 0.1,
+  "eos_token_id": 1,
+  "feed_forward_proj": "gated-gelu",
+  "initializer_factor": 1.0,
+  "is_encoder_decoder": true,
+  "layer_norm_epsilon": 1e-06,
+  "model_type": "mt5",
+  "num_decoder_layers": 12,
+  "num_heads": 12,
+  "num_layers": 12,
+  "output_past": true,
+  "pad_token_id": 0,
+  "relative_attention_num_buckets": 32,
+  "tie_word_embeddings": false,
+  "tokenizer_class": "T5Tokenizer",
+  "torch_dtype": "float32",
+  "transformers_version": "4.11.3",
+  "use_cache": true,
+  "vocab_size": 250100
+}

pytorch_model.bin

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:18b608a7836e1e81f15b72d9bbd255919547dbb21b035c9cf0abcad95e4fc8f2
+size 2329633625

special_tokens_map.json

@@ -0,0 +1 @@
+{"eos_token": "</s>", "unk_token": "<unk>", "pad_token": "<pad>"}

spiece.model

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:ef78f86560d809067d12bac6c09f19a462cb3af3f54d2b8acbba26e1433125d6
+size 4309802

tokenizer_config.json

@@ -0,0 +1 @@
+{"eos_token": "</s>", "unk_token": "<unk>", "pad_token": "<pad>", "extra_ids": 0, "additional_special_tokens": null, "special_tokens_map_file": "/home/patrick/.cache/torch/transformers/685ac0ca8568ec593a48b61b0a3c272beee9bc194a3c7241d15dcadb5f875e53.f76030f3ec1b96a8199b2593390c610e76ca8028ef3d24680000619ffb646276", "name_or_path": "google/mt5-base", "sp_model_kwargs": {}, "tokenizer_class": "T5Tokenizer"}