Training in progress, step 1000

config.json

@@ -1,5 +1,5 @@
 {
-  "_name_or_path": "openai/whisper-tiny",
+  "_name_or_path": "openai/whisper-medium",
   "activation_dropout": 0.0,
   "activation_function": "gelu",
   "architectures": [
@@ -11,17 +11,17 @@
     50257
   ],
   "bos_token_id": 50257,
-  "d_model": 384,
-  "decoder_attention_heads": 6,
-  "decoder_ffn_dim": 1536,
+  "d_model": 1024,
+  "decoder_attention_heads": 16,
+  "decoder_ffn_dim": 4096,
   "decoder_layerdrop": 0.0,
-  "decoder_layers": 4,
+  "decoder_layers": 24,
   "decoder_start_token_id": 50258,
   "dropout": 0.0,
-  "encoder_attention_heads": 6,
-  "encoder_ffn_dim": 1536,
+  "encoder_attention_heads": 16,
+  "encoder_ffn_dim": 4096,
   "encoder_layerdrop": 0.0,
-  "encoder_layers": 4,
+  "encoder_layers": 24,
   "eos_token_id": 50257,
   "forced_decoder_ids": null,
   "init_std": 0.02,
@@ -30,7 +30,7 @@
   "max_source_positions": 1500,
   "max_target_positions": 448,
   "model_type": "whisper",
-  "num_hidden_layers": 4,
+  "num_hidden_layers": 24,
   "num_mel_bins": 80,
   "pad_token_id": 50257,
   "scale_embedding": false,

pytorch_model.bin

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:0d7fb56b1521b516ee5e2864fd4e40fa894bd6de2e431ea5d5312888f79aa5aa
-size 151098921
+oid sha256:094fe63aeb332c29b588288a0a989111394ac5cb1ff02ebe44f4d558a2a12b41
+size 3055754841

tokenizer_config.json

@@ -19,7 +19,7 @@
   },
   "errors": "replace",
   "model_max_length": 1024,
-  "name_or_path": "openai/whisper-tiny",
+  "name_or_path": "openai/whisper-medium",
   "pad_token": null,
   "processor_class": "WhisperProcessor",
   "return_attention_mask": false,

training_args.bin

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:46661f32574b36ae098957883106754632cf269d942316b421fc00ad630f4a9b
+oid sha256:6a24a354d10cbc19da5b395016c86aca67ce02fbab4986656ecdcc7488a8efb7
 size 3707