Training in progress, step 10
Browse files- config.json +196 -0
- generation_config.json +8 -0
- model.safetensors +3 -0
- training_args.bin +3 -0
config.json
ADDED
|
@@ -0,0 +1,196 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"architectures": [
|
| 3 |
+
"ASRModel"
|
| 4 |
+
],
|
| 5 |
+
"attn_implementation": "flash_attention_2",
|
| 6 |
+
"audio_config": {
|
| 7 |
+
"_name_or_path": "openai/whisper-large-v3-turbo",
|
| 8 |
+
"activation_dropout": 0.0,
|
| 9 |
+
"activation_function": "gelu",
|
| 10 |
+
"apply_spec_augment": false,
|
| 11 |
+
"architectures": [
|
| 12 |
+
"WhisperForConditionalGeneration"
|
| 13 |
+
],
|
| 14 |
+
"attention_dropout": 0.0,
|
| 15 |
+
"bos_token_id": 50257,
|
| 16 |
+
"classifier_proj_size": 256,
|
| 17 |
+
"d_model": 1280,
|
| 18 |
+
"decoder_attention_heads": 20,
|
| 19 |
+
"decoder_ffn_dim": 5120,
|
| 20 |
+
"decoder_layerdrop": 0.0,
|
| 21 |
+
"decoder_layers": 4,
|
| 22 |
+
"decoder_start_token_id": 50258,
|
| 23 |
+
"dropout": 0.0,
|
| 24 |
+
"dtype": "float16",
|
| 25 |
+
"encoder_attention_heads": 20,
|
| 26 |
+
"encoder_ffn_dim": 5120,
|
| 27 |
+
"encoder_layerdrop": 0.0,
|
| 28 |
+
"encoder_layers": 32,
|
| 29 |
+
"eos_token_id": 50257,
|
| 30 |
+
"init_std": 0.02,
|
| 31 |
+
"mask_feature_length": 10,
|
| 32 |
+
"mask_feature_min_masks": 0,
|
| 33 |
+
"mask_feature_prob": 0.0,
|
| 34 |
+
"mask_time_length": 10,
|
| 35 |
+
"mask_time_min_masks": 2,
|
| 36 |
+
"mask_time_prob": 0.05,
|
| 37 |
+
"max_source_positions": 1500,
|
| 38 |
+
"max_target_positions": 448,
|
| 39 |
+
"median_filter_width": 7,
|
| 40 |
+
"model_type": "whisper",
|
| 41 |
+
"num_hidden_layers": 32,
|
| 42 |
+
"num_mel_bins": 128,
|
| 43 |
+
"pad_token_id": 50257,
|
| 44 |
+
"scale_embedding": false,
|
| 45 |
+
"use_cache": true,
|
| 46 |
+
"use_weighted_layer_sum": false,
|
| 47 |
+
"vocab_size": 51866
|
| 48 |
+
},
|
| 49 |
+
"audio_downsample_rate": 5,
|
| 50 |
+
"audio_model_id": "openai/whisper-large-v3-turbo",
|
| 51 |
+
"audio_sample_rate": 16000,
|
| 52 |
+
"auto_map": {
|
| 53 |
+
"AutoConfig": "asr_config.ASRConfig",
|
| 54 |
+
"AutoModel": "asr_modeling.ASRModel",
|
| 55 |
+
"AutoModelForSpeechSeq2Seq": "asr_modeling.ASRModel",
|
| 56 |
+
"AutoProcessor": "asr_processing.ASRProcessor"
|
| 57 |
+
},
|
| 58 |
+
"custom_pipelines": {
|
| 59 |
+
"automatic-speech-recognition": {
|
| 60 |
+
"impl": "asr_pipeline.ASRPipeline",
|
| 61 |
+
"pt": [
|
| 62 |
+
"AutoModelForSpeechSeq2Seq"
|
| 63 |
+
],
|
| 64 |
+
"tf": [],
|
| 65 |
+
"type": "audio"
|
| 66 |
+
}
|
| 67 |
+
},
|
| 68 |
+
"dtype": "bfloat16",
|
| 69 |
+
"encoder_dim": 1280,
|
| 70 |
+
"inference_diversity_penalty": 0.0,
|
| 71 |
+
"inference_warmup_tokens": 10,
|
| 72 |
+
"llm_dim": 2048,
|
| 73 |
+
"max_new_tokens": 128,
|
| 74 |
+
"min_new_tokens": 1,
|
| 75 |
+
"model_dtype": "bfloat16",
|
| 76 |
+
"model_type": "asr_model",
|
| 77 |
+
"pipeline_tag": "automatic-speech-recognition",
|
| 78 |
+
"projector_dropout": 0.1,
|
| 79 |
+
"projector_hidden_dim": 5120,
|
| 80 |
+
"projector_init_std": 0.02,
|
| 81 |
+
"projector_pool_stride": 2,
|
| 82 |
+
"repetition_penalty": 1.05,
|
| 83 |
+
"system_prompt": "/no_think /system_override",
|
| 84 |
+
"text_config": {
|
| 85 |
+
"_name_or_path": "HuggingFaceTB/SmolLM3-3B",
|
| 86 |
+
"architectures": [
|
| 87 |
+
"SmolLM3ForCausalLM"
|
| 88 |
+
],
|
| 89 |
+
"attention_bias": false,
|
| 90 |
+
"attention_dropout": 0.0,
|
| 91 |
+
"bos_token_id": null,
|
| 92 |
+
"dtype": "bfloat16",
|
| 93 |
+
"eos_token_id": 128012,
|
| 94 |
+
"hidden_act": "silu",
|
| 95 |
+
"hidden_size": 2048,
|
| 96 |
+
"initializer_range": 0.02,
|
| 97 |
+
"intermediate_size": 11008,
|
| 98 |
+
"layer_types": [
|
| 99 |
+
"full_attention",
|
| 100 |
+
"full_attention",
|
| 101 |
+
"full_attention",
|
| 102 |
+
"full_attention",
|
| 103 |
+
"full_attention",
|
| 104 |
+
"full_attention",
|
| 105 |
+
"full_attention",
|
| 106 |
+
"full_attention",
|
| 107 |
+
"full_attention",
|
| 108 |
+
"full_attention",
|
| 109 |
+
"full_attention",
|
| 110 |
+
"full_attention",
|
| 111 |
+
"full_attention",
|
| 112 |
+
"full_attention",
|
| 113 |
+
"full_attention",
|
| 114 |
+
"full_attention",
|
| 115 |
+
"full_attention",
|
| 116 |
+
"full_attention",
|
| 117 |
+
"full_attention",
|
| 118 |
+
"full_attention",
|
| 119 |
+
"full_attention",
|
| 120 |
+
"full_attention",
|
| 121 |
+
"full_attention",
|
| 122 |
+
"full_attention",
|
| 123 |
+
"full_attention",
|
| 124 |
+
"full_attention",
|
| 125 |
+
"full_attention",
|
| 126 |
+
"full_attention",
|
| 127 |
+
"full_attention",
|
| 128 |
+
"full_attention",
|
| 129 |
+
"full_attention",
|
| 130 |
+
"full_attention",
|
| 131 |
+
"full_attention",
|
| 132 |
+
"full_attention",
|
| 133 |
+
"full_attention",
|
| 134 |
+
"full_attention"
|
| 135 |
+
],
|
| 136 |
+
"max_position_embeddings": 65536,
|
| 137 |
+
"max_window_layers": 28,
|
| 138 |
+
"mlp_bias": false,
|
| 139 |
+
"model_type": "smollm3",
|
| 140 |
+
"no_rope_layer_interval": 4,
|
| 141 |
+
"no_rope_layers": [
|
| 142 |
+
1,
|
| 143 |
+
1,
|
| 144 |
+
1,
|
| 145 |
+
0,
|
| 146 |
+
1,
|
| 147 |
+
1,
|
| 148 |
+
1,
|
| 149 |
+
0,
|
| 150 |
+
1,
|
| 151 |
+
1,
|
| 152 |
+
1,
|
| 153 |
+
0,
|
| 154 |
+
1,
|
| 155 |
+
1,
|
| 156 |
+
1,
|
| 157 |
+
0,
|
| 158 |
+
1,
|
| 159 |
+
1,
|
| 160 |
+
1,
|
| 161 |
+
0,
|
| 162 |
+
1,
|
| 163 |
+
1,
|
| 164 |
+
1,
|
| 165 |
+
0,
|
| 166 |
+
1,
|
| 167 |
+
1,
|
| 168 |
+
1,
|
| 169 |
+
0,
|
| 170 |
+
1,
|
| 171 |
+
1,
|
| 172 |
+
1,
|
| 173 |
+
0,
|
| 174 |
+
1,
|
| 175 |
+
1,
|
| 176 |
+
1,
|
| 177 |
+
0
|
| 178 |
+
],
|
| 179 |
+
"num_attention_heads": 16,
|
| 180 |
+
"num_hidden_layers": 36,
|
| 181 |
+
"num_key_value_heads": 4,
|
| 182 |
+
"pretraining_tp": 2,
|
| 183 |
+
"rms_norm_eps": 1e-06,
|
| 184 |
+
"rope_scaling": null,
|
| 185 |
+
"rope_theta": 5000000.0,
|
| 186 |
+
"sliding_window": null,
|
| 187 |
+
"use_cache": false,
|
| 188 |
+
"use_sliding_window": false,
|
| 189 |
+
"vocab_size": 128257
|
| 190 |
+
},
|
| 191 |
+
"text_model_id": "HuggingFaceTB/SmolLM3-3B",
|
| 192 |
+
"transformers_version": "4.57.1",
|
| 193 |
+
"use_cache": false,
|
| 194 |
+
"user_prompt": "Transcribe: <audio>",
|
| 195 |
+
"vocab_size": 128257
|
| 196 |
+
}
|
generation_config.json
ADDED
|
@@ -0,0 +1,8 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"do_sample": true,
|
| 3 |
+
"eos_token_id": 128012,
|
| 4 |
+
"pad_token_id": 128004,
|
| 5 |
+
"temperature": 0.6,
|
| 6 |
+
"top_p": 0.95,
|
| 7 |
+
"transformers_version": "4.57.1"
|
| 8 |
+
}
|
model.safetensors
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:a76bfeb0e8ecc9560f1a2fd66de639e037d2294f324aa82a5b11a889555d84b4
|
| 3 |
+
size 73400656
|
training_args.bin
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:8d27d3de3c6a0e78a87221a11b4af8235c7576dfbf2a9e7eb6ca391f8bfbf16e
|
| 3 |
+
size 5905
|