End of training

README.md

@@ -25,7 +25,7 @@ model-index:
     metrics:
     - name: Wer
       type: wer
-      value: 30.810810810810814
+      value: 2.608695652173913
 ---
 
 <!-- This model card has been generated automatically according to the information the Trainer had access to. You
@@ -36,7 +36,7 @@ should probably proofread and complete it, then remove this comment. -->
 This model is a fine-tuned version of [openai/whisper-base](https://huggingface.co/openai/whisper-base) on the Domain Based voice dataset.
 It achieves the following results on the evaluation set:
 - Loss: 0.0000
-- Wer: 30.8108
+- Wer: 2.6087
 
 ## Model description
 
@@ -55,24 +55,26 @@ More information needed
 ### Training hyperparameters
 
 The following hyperparameters were used during training:
-- learning_rate: 1e-05
-- train_batch_size: 16
-- eval_batch_size: 8
+- learning_rate: 5e-06
+- train_batch_size: 4
+- eval_batch_size: 4
 - seed: 42
+- gradient_accumulation_steps: 4
+- total_train_batch_size: 16
 - optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
 - lr_scheduler_type: linear
-- lr_scheduler_warmup_steps: 500
-- training_steps: 4000
+- lr_scheduler_warmup_steps: 100
+- training_steps: 2000
 - mixed_precision_training: Native AMP
 
 ### Training results
 
-| Training Loss | Epoch    | Step | Validation Loss | Wer     |
-|:-------------:|:--------:|:----:|:---------------:|:-------:|
-| 0.0           | 142.8571 | 1000 | 0.0000          | 61.2613 |
-| 0.0           | 285.7143 | 2000 | 0.0000          | 38.1982 |
-| 0.0           | 428.5714 | 3000 | 0.0000          | 74.2342 |
-| 0.0           | 571.4286 | 4000 | 0.0000          | 30.8108 |
+| Training Loss | Epoch | Step | Validation Loss | Wer    |
+|:-------------:|:-----:|:----:|:---------------:|:------:|
+| 0.0001        | 100.0 | 500  | 0.0001          | 1.7391 |
+| 0.0           | 200.0 | 1000 | 0.0001          | 1.7391 |
+| 0.0           | 300.0 | 1500 | 0.0000          | 2.6087 |
+| 0.0           | 400.0 | 2000 | 0.0000          | 2.6087 |
 
 
 ### Framework versions

generation_config.json

@@ -142,6 +142,7 @@
     "<|yo|>": 50325,
     "<|zh|>": 50260
   },
+  "language": "english",
   "max_initial_timestamp_index": 50,
   "max_length": 448,
   "no_timestamps_token_id": 50363,

model.safetensors

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:261778951a1304f991f4a0a6fb4190e1c4cd57375f2a1f46565a3a03b4533d6b
+oid sha256:a53da9f5b2deaf5b384d54b1bb88e19e50a2d051a4c9100121be634eadc858cf
 size 290403936