update model

config.json

@@ -1,7 +1,8 @@
 {
-  "_name_or_path": "openai/whisper-small",
+  "_name_or_path": "whisper-th-small-aug",
   "activation_dropout": 0.0,
   "activation_function": "gelu",
+  "apply_spec_augment": true,
   "architectures": [
     "WhisperForConditionalGeneration"
   ],
@@ -11,6 +12,7 @@
     50257
   ],
   "bos_token_id": 50257,
+  "classifier_proj_size": 256,
   "d_model": 768,
   "decoder_attention_heads": 12,
   "decoder_ffn_dim": 3072,
@@ -26,16 +28,24 @@
   "forced_decoder_ids": null,
   "init_std": 0.02,
   "is_encoder_decoder": true,
+  "mask_feature_length": 64,
+  "mask_feature_min_masks": 0,
+  "mask_feature_prob": 0.3,
+  "mask_time_length": 10,
+  "mask_time_min_masks": 2,
+  "mask_time_prob": 0.3,
   "max_length": 448,
   "max_source_positions": 1500,
   "max_target_positions": 448,
+  "median_filter_width": 7,
   "model_type": "whisper",
   "num_hidden_layers": 12,
   "num_mel_bins": 80,
   "pad_token_id": 50257,
   "scale_embedding": false,
-  "torch_dtype": "float16",
-  "transformers_version": "4.27.0.dev0",
-  "use_cache": false,
+  "torch_dtype": "float32",
+  "transformers_version": "4.31.0.dev0",
+  "use_cache": true,
+  "use_weighted_layer_sum": false,
   "vocab_size": 51865
 }

generation_config.json

@@ -14,10 +14,6 @@
     [
       2,
       50359
-    ],
-    [
-      3,
-      50363
     ]
   ],
   "is_multilingual": true,
@@ -125,7 +121,7 @@
   "max_initial_timestamp_index": 1,
   "max_length": 448,
   "no_timestamps_token_id": 50363,
-  "pad_token_id": 50256,
+  "pad_token_id": 50257,
   "return_timestamps": false,
   "suppress_tokens": [
     1,
@@ -211,6 +207,8 @@
     49870,
     50254,
     50258,
+    50358,
+    50359,
     50360,
     50361,
     50362
@@ -219,5 +217,5 @@
     "transcribe": 50359,
     "translate": 50358
   },
-  "transformers_version": "4.27.0.dev0"
+  "transformers_version": "4.31.0.dev0"
 }

pytorch_model.bin

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:dc8d8ce98f7b8ed8ae3fe0154f1911f137be21ce23cfaa8fece1f7a80bd8aaf1
-size 483536061
+oid sha256:6c48b903002d62eddd6bdac0af1fcb9d1825f7616b03562dbceaa0a029943206
+size 967102926

training_args.bin

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:c2fb67b34d1aaaf327e038dd1e7ff6fe1ac4201d6dc60f7bcbc9a0a6b743801e
-size 4731
+oid sha256:92e0d28a477d6fbed86250bc2bdb7420c074fb1d27d9be486675bd887280e570
+size 4352