Samuael/amBART

README.md

@@ -1,5 +1,5 @@
 ---
-base_model: Samuael/amBART_1000
+base_model: Samuael/amBART
 tags:
 - generated_from_trainer
 model-index:
@@ -12,17 +12,17 @@ should probably proofread and complete it, then remove this comment. -->
 
 # amBART
 
-This model is a fine-tuned version of [Samuael/amBART_1000](https://huggingface.co/Samuael/amBART_1000) on an unknown dataset.
+This model is a fine-tuned version of [Samuael/amBART](https://huggingface.co/Samuael/amBART) on an unknown dataset.
 It achieves the following results on the evaluation set:
-- eval_loss: 2.9935
-- eval_wer: 0.9817
-- eval_cer: 0.7707
-- eval_bleu: 0.7418
-- eval_runtime: 9.129
-- eval_samples_per_second: 39.325
-- eval_steps_per_second: 0.329
-- epoch: 3.0
-- step: 2127
+- eval_loss: 3.4340
+- eval_wer: 0.9966
+- eval_cer: 0.8055
+- eval_bleu: 0.1833
+- eval_runtime: 8.1414
+- eval_samples_per_second: 44.096
+- eval_steps_per_second: 0.246
+- epoch: 11.0
+- step: 3905
 
 ## Model description
 
@@ -41,14 +41,13 @@ More information needed
 ### Training hyperparameters
 
 The following hyperparameters were used during training:
-- learning_rate: 0.002
-- train_batch_size: 128
-- eval_batch_size: 128
+- learning_rate: 0.0002
+- train_batch_size: 256
+- eval_batch_size: 256
 - seed: 42
 - optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
 - lr_scheduler_type: linear
 - num_epochs: 50
-- mixed_precision_training: Native AMP
 
 ### Framework versions
 

config.json

@@ -1,5 +1,5 @@
 {
-  "_name_or_path": "Samuael/amBART_1000",
+  "_name_or_path": "Samuael/amBART",
   "activation_dropout": 0.0,
   "activation_function": "gelu",
   "architectures": [

model.safetensors

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:1504bcd0758ac82ab09f78650ebc0c7c431edde67e78c5fb92b4befcf30c490e
+oid sha256:e48b1ce78fe21170b4789afb20904bfd8f3b41f68909381b4fb2c806011a1255
 size 180813204

training_args.bin

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:f19be77b3b5f0472323533764679e1f13184f30ee05c97f781d3b916fe948fbe
+oid sha256:89e3c4eea5f7f0e97067c89e15fc1896956960907cf92ea9ab7b5f40a6a93afe
 size 4984