End of training

README.md

@@ -1,42 +1,40 @@
 ---
-language:
-- dv
 license: apache-2.0
-base_model: openai/whisper-small
+base_model: openai/whisper-tiny
 tags:
 - generated_from_trainer
 datasets:
-- mozilla-foundation/common_voice_13_0
+- PolyAI/minds14
 metrics:
 - wer
 model-index:
-- name: Whisper-Small-Dv-fine-tuned
+- name: whisper-small-dv
   results:
   - task:
       name: Automatic Speech Recognition
       type: automatic-speech-recognition
     dataset:
-      name: Common Voice 13
-      type: mozilla-foundation/common_voice_13_0
-      config: dv
-      split: test
-      args: dv
+      name: PolyAI/minds14
+      type: PolyAI/minds14
+      config: en-US
+      split: train
+      args: en-US
     metrics:
     - name: Wer
       type: wer
-      value: 12.621274820043816
+      value: 35.6091030789826
 ---
 
 <!-- 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. -->
 
-# Whisper-Small-Dv-fine-tuned
+# whisper-small-dv
 
-This model is a fine-tuned version of [openai/whisper-small](https://huggingface.co/openai/whisper-small) on the Common Voice 13 dataset.
+This model is a fine-tuned version of [openai/whisper-tiny](https://huggingface.co/openai/whisper-tiny) on the PolyAI/minds14 dataset.
 It achieves the following results on the evaluation set:
-- Loss: 0.1648
-- Wer Ortho: 60.1574
-- Wer: 12.6213
+- Loss: 0.7160
+- Wer Ortho: 36.2369
+- Wer: 35.6091
 
 ## Model description
 
@@ -56,20 +54,23 @@ More information needed
 
 The following hyperparameters were used during training:
 - learning_rate: 1e-05
-- train_batch_size: 32
-- eval_batch_size: 32
+- train_batch_size: 8
+- eval_batch_size: 8
 - seed: 42
+- gradient_accumulation_steps: 4
+- total_train_batch_size: 32
 - optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
 - lr_scheduler_type: constant_with_warmup
 - lr_scheduler_warmup_steps: 50
-- training_steps: 500
+- training_steps: 200
 - mixed_precision_training: Native AMP
 
 ### Training results
 
-| Training Loss | Epoch  | Step | Validation Loss | Wer Ortho | Wer     |
-|:-------------:|:------:|:----:|:---------------:|:---------:|:-------:|
-| 0.0698        | 3.2468 | 500  | 0.1648          | 60.1574   | 12.6213 |
+| Training Loss | Epoch   | Step | Validation Loss | Wer Ortho | Wer     |
+|:-------------:|:-------:|:----:|:---------------:|:---------:|:-------:|
+| 0.2296        | 7.1429  | 100  | 0.5760          | 34.1463   | 33.7349 |
+| 0.0048        | 14.2857 | 200  | 0.7160          | 36.2369   | 35.6091 |
 
 
 ### Framework versions

config.json

@@ -1,5 +1,5 @@
 {
-  "_name_or_path": "openai/whisper-small",
+  "_name_or_path": "openai/whisper-tiny",
   "activation_dropout": 0.0,
   "activation_function": "gelu",
   "apply_spec_augment": false,
@@ -13,17 +13,17 @@
   ],
   "bos_token_id": 50257,
   "classifier_proj_size": 256,
-  "d_model": 768,
-  "decoder_attention_heads": 12,
-  "decoder_ffn_dim": 3072,
+  "d_model": 384,
+  "decoder_attention_heads": 6,
+  "decoder_ffn_dim": 1536,
   "decoder_layerdrop": 0.0,
-  "decoder_layers": 12,
+  "decoder_layers": 4,
   "decoder_start_token_id": 50258,
   "dropout": 0.0,
-  "encoder_attention_heads": 12,
-  "encoder_ffn_dim": 3072,
+  "encoder_attention_heads": 6,
+  "encoder_ffn_dim": 1536,
   "encoder_layerdrop": 0.0,
-  "encoder_layers": 12,
+  "encoder_layers": 4,
   "eos_token_id": 50257,
   "forced_decoder_ids": [
     [
@@ -52,7 +52,7 @@
   "max_target_positions": 448,
   "median_filter_width": 7,
   "model_type": "whisper",
-  "num_hidden_layers": 12,
+  "num_hidden_layers": 4,
   "num_mel_bins": 80,
   "pad_token_id": 50257,
   "scale_embedding": false,
@@ -140,6 +140,8 @@
     49870,
     50254,
     50258,
+    50358,
+    50359,
     50360,
     50361,
     50362

generation_config.json

@@ -1,43 +1,27 @@
 {
   "alignment_heads": [
     [
-      5,
-      3
-    ],
-    [
-      5,
-      9
+      2,
+      2
     ],
     [
-      8,
+      3,
       0
     ],
     [
-      8,
-      4
-    ],
-    [
-      8,
-      7
+      3,
+      2
     ],
     [
-      8,
-      8
-    ],
-    [
-      9,
-      0
-    ],
-    [
-      9,
-      7
+      3,
+      3
     ],
     [
-      9,
-      9
+      3,
+      4
     ],
     [
-      10,
+      3,
       5
     ]
   ],

model.safetensors

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:927fbd1f5fd870a3b83268986757002a921320d684c64a212def802a11cacdb1
-size 966995080
+oid sha256:e402f541632d00e2956790750390c78a6094c78d742b62b91187f976d0862c2e
+size 151061672

training_args.bin

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:161cad3ce9212d9981716768b0f7cf7d7baeded183afbddebf680b12dd26e8ad
+oid sha256:a58d979b65d06f10ff7201a06c65afc6270fd97fc5e055e99a8af7a35d67c48f
 size 5304