Training complete

README.md

@@ -1,14 +1,29 @@
 ---
 license: mit
-base_model: facebook/m2m100_418M
+base_model: kazandaev/m2m100_418M
 tags:
 - translation
 - generated_from_trainer
 datasets:
 - wmt16
+metrics:
+- bleu
 model-index:
 - name: m2m100_418M
-  results: []
+  results:
+  - task:
+      name: Sequence-to-sequence Language Modeling
+      type: text2text-generation
+    dataset:
+      name: wmt16
+      type: wmt16
+      config: ru-en
+      split: validation
+      args: ru-en
+    metrics:
+    - name: Bleu
+      type: bleu
+      value: 32.0585
 ---
 
 <!-- This model card has been generated automatically according to the information the Trainer had access to. You
@@ -16,7 +31,11 @@ should probably proofread and complete it, then remove this comment. -->
 
 # m2m100_418M
 
-This model is a fine-tuned version of [facebook/m2m100_418M](https://huggingface.co/facebook/m2m100_418M) on the wmt16 dataset.
+This model is a fine-tuned version of [kazandaev/m2m100_418M](https://huggingface.co/kazandaev/m2m100_418M) on the wmt16 dataset.
+It achieves the following results on the evaluation set:
+- Loss: 0.8954
+- Bleu: 32.0585
+- Gen Len: 36.1643
 
 ## Model description
 
@@ -36,14 +55,22 @@ More information needed
 
 The following hyperparameters were used during training:
 - learning_rate: 2e-05
-- train_batch_size: 3
+- train_batch_size: 4
 - eval_batch_size: 2
 - seed: 42
 - gradient_accumulation_steps: 10
-- total_train_batch_size: 30
+- total_train_batch_size: 40
 - optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
 - lr_scheduler_type: linear
-- num_epochs: 1
+- num_epochs: 2
+
+### Training results
+
+| Training Loss | Epoch | Step  | Validation Loss | Bleu    | Gen Len |
+|:-------------:|:-----:|:-----:|:---------------:|:-------:|:-------:|
+| 0.8087        | 1.0   | 47790 | 0.9542          | 30.786  | 36.1469 |
+| 0.7266        | 2.0   | 95580 | 0.8954          | 32.0585 | 36.1643 |
+
 
 ### Framework versions