cucl2/AnyAudio-Judge-7B
063
1{2 "architectures": [3 "Qwen2_5OmniForConditionalGeneration"4 ],5 "dtype": "bfloat16",6 "enable_audio_output": false,7 "enable_talker": true,8 "eos_token_id": 151645,9 "hidden_size": 3584,10 "keys_to_ignore_at_inference": [11 "past_key_values",12 "hidden_states",13 "attention_mask"14 ],15 "model_type": "qwen2_5_omni",16 "pad_token_id": 151643,17 "talker_config": {18 "_attn_implementation_autoset": true,19 "_name_or_path": "Qwen2.5-Omni-7B/talker",20 "architectures": [21 "Qwen2OmniTalkerForConditionalGeneration"22 ],23 "attention_dropout": 0.0,24 "audio_end_token_id": 151648,25 "audio_start_token_id": 151647,26 "audio_token_index": 151646,27 "dtype": "bfloat16",28 "embedding_size": 3584,29 "head_dim": 128,30 "hidden_act": "silu",31 "hidden_size": 896,32 "image_token_index": 151655,33 "init_std": 0.02,34 "initializer_range": 0.02,35 "intermediate_size": 18944,36 "layer_types": [37 "full_attention",38 "full_attention",39 "full_attention",40 "full_attention",41 "full_attention",42 "full_attention",43 "full_attention",44 "full_attention",45 "full_attention",46 "full_attention",47 "full_attention",48 "full_attention",49 "full_attention",50 "full_attention",51 "full_attention",52 "full_attention",53 "full_attention",54 "full_attention",55 "full_attention",56 "full_attention",57 "full_attention",58 "full_attention",59 "full_attention",60 "full_attention"61 ],62 "max_position_embeddings": 32768,63 "max_window_layers": 28,64 "model_type": "qwen2_5_omni_talker",65 "num_attention_heads": 12,66 "num_hidden_layers": 24,67 "num_key_value_heads": 4,68 "position_id_per_seconds": 25,69 "rms_norm_eps": 1e-06,70 "rope_scaling": {71 "mrope_section": [72 16,73 24,74 2475 ],76 "rope_type": "default",77 "type": "default"78 },79 "rope_theta": 1000000.0,80 "seconds_per_chunk": 2,81 "sliding_window": null,82 "spatial_merge_size": 2,83 "tts_codec_end_token_id": 8294,84 "tts_codec_mask_token_id": 8296,85 "tts_codec_pad_token_id": 8292,86 "tts_codec_start_token_id": 8293,87 "tts_text_end_token_id": 151861,88 "tts_text_pad_token_id": 151859,89 "tts_text_start_token_id": 151860,90 "use_cache": true,91 "use_sliding_window": false,92 "video_token_index": 151656,93 "vision_end_token_id": 151653,94 "vision_start_token_id": 151652,95 "vocab_size": 844896 },97 "thinker_config": {98 "_attn_implementation_autoset": true,99 "_name_or_path": "Qwen2.5-Omni-7B/thinker",100 "architectures": [101 "Qwen2OmniNaViTThinkerForConditionalGeneration"102 ],103 "audio_config": {104 "_attn_implementation_autoset": true,105 "_name_or_path": "",106 "activation_dropout": 0.0,107 "activation_function": "gelu",108 "add_cross_attention": false,109 "architectures": null,110 "attention_dropout": 0.0,111 "bad_words_ids": null,112 "begin_suppress_tokens": null,113 "bos_token_id": null,114 "chunk_size_feed_forward": 0,115 "cross_attention_hidden_size": null,116 "d_model": 1280,117 "decoder_start_token_id": null,118 "diversity_penalty": 0.0,119 "do_sample": false,120 "dropout": 0.0,121 "dtype": "bfloat16",122 "early_stopping": false,123 "encoder_attention_heads": 20,124 "encoder_ffn_dim": 5120,125 "encoder_layerdrop": 0.0,126 "encoder_layers": 32,127 "encoder_no_repeat_ngram_size": 0,128 "eos_token_id": null,129 "exponential_decay_length_penalty": null,130 "finetuning_task": null,131 "forced_bos_token_id": null,132 "forced_eos_token_id": null,133 "id2label": {134 "0": "LABEL_0",135 "1": "LABEL_1"136 },137 "init_std": 0.02,138 "initializer_range": 0.02,139 "is_decoder": false,140 "is_encoder_decoder": false,141 "label2id": {142 "LABEL_0": 0,143 "LABEL_1": 1144 },145 "length_penalty": 1.0,146 "max_length": 20,147 "max_source_positions": 1500,148 "min_length": 0,149 "model_type": "qwen2_5_omni_audio_encoder",150 "n_window": 100,151 "no_repeat_ngram_size": 0,152 "num_beam_groups": 1,153 "num_beams": 1,154 "num_hidden_layers": 32,155 "num_mel_bins": 128,156 "num_return_sequences": 1,157 "output_attentions": false,158 "output_dim": 3584,159 "output_hidden_states": false,160 "output_scores": false,161 "pad_token_id": null,162 "prefix": null,163 "problem_type": null,164 "pruned_heads": {},165 "remove_invalid_values": false,166 "repetition_penalty": 1.0,167 "return_dict": true,168 "return_dict_in_generate": false,169 "scale_embedding": false,170 "sep_token_id": null,171 "suppress_tokens": null,172 "task_specific_params": null,173 "temperature": 1.0,174 "tf_legacy_loss": false,175 "tie_encoder_decoder": false,176 "tie_word_embeddings": true,177 "tokenizer_class": null,178 "top_k": 50,179 "top_p": 1.0,180 "torchscript": false,181 "typical_p": 1.0,182 "use_bfloat16": false183 },184 "audio_end_token_id": 151648,185 "audio_start_token_id": 151647,186 "bos_token_id": 151644,187 "dtype": "bfloat16",188 "eos_token_id": 151645,189 "ignore_index": -100,190 "init_std": 0.02,191 "initializer_range": 0.02,192 "model_type": "qwen2_5_omni_thinker",193 "pad_token_id": 151643,194 "position_id_per_seconds": 25,195 "seconds_per_chunk": 2,196 "text_config": {197 "_name_or_path": "",198 "add_cross_attention": false,199 "architectures": null,200 "attention_dropout": 0.0,201 "bad_words_ids": null,202 "begin_suppress_tokens": null,203 "bos_token_id": null,204 "chunk_size_feed_forward": 0,205 "cross_attention_hidden_size": null,206 "decoder_start_token_id": null,207 "diversity_penalty": 0.0,208 "do_sample": false,209 "dtype": "bfloat16",210 "early_stopping": false,211 "encoder_no_repeat_ngram_size": 0,212 "eos_token_id": null,213 "exponential_decay_length_penalty": null,214 "finetuning_task": null,215 "forced_bos_token_id": null,216 "forced_eos_token_id": null,217 "hidden_act": "silu",218 "hidden_size": 3584,219 "id2label": {220 "0": "LABEL_0",221 "1": "LABEL_1"222 },223 "init_std": 0.02,224 "initializer_range": 0.02,225 "intermediate_size": 18944,226 "is_decoder": false,227 "is_encoder_decoder": false,228 "label2id": {229 "LABEL_0": 0,230 "LABEL_1": 1231 },232 "layer_types": [233 "full_attention",234 "full_attention",235 "full_attention",236 "full_attention",237 "full_attention",238 "full_attention",239 "full_attention",240 "full_attention",241 "full_attention",242 "full_attention",243 "full_attention",244 "full_attention",245 "full_attention",246 "full_attention",247 "full_attention",248 "full_attention",249 "full_attention",250 "full_attention",251 "full_attention",252 "full_attention",253 "full_attention",254 "full_attention",255 "full_attention",256 "full_attention",257 "full_attention",258 "full_attention",259 "full_attention",260 "full_attention"261 ],262 "length_penalty": 1.0,263 "max_length": 20,264 "max_position_embeddings": 32768,265 "max_window_layers": 28,266 "min_length": 0,267 "model_type": "qwen2_5_omni_text",268 "no_repeat_ngram_size": 0,269 "num_attention_heads": 28,270 "num_beam_groups": 1,271 "num_beams": 1,272 "num_hidden_layers": 28,273 "num_key_value_heads": 4,274 "num_return_sequences": 1,275 "output_attentions": false,276 "output_hidden_states": false,277 "output_scores": false,278 "pad_token_id": 151643,279 "prefix": null,280 "problem_type": null,281 "pruned_heads": {},282 "remove_invalid_values": false,283 "repetition_penalty": 1.0,284 "return_dict": true,285 "return_dict_in_generate": false,286 "rms_norm_eps": 1e-06,287 "rope_scaling": {288 "mrope_section": [289 16,290 24,291 24292 ],293 "rope_type": "default",294 "type": "default"295 },296 "rope_theta": 1000000.0,297 "sep_token_id": null,298 "sliding_window": null,299 "suppress_tokens": null,300 "task_specific_params": null,301 "temperature": 1.0,302 "tie_encoder_decoder": false,303 "tie_word_embeddings": false,304 "tokenizer_class": null,305 "top_k": 50,306 "top_p": 1.0,307 "torchscript": false,308 "typical_p": 1.0,309 "use_cache": false,310 "use_sliding_window": false,311 "vocab_size": 152064312 },313 "user_token_id": 872,314 "vision_config": {315 "_attn_implementation_autoset": true,316 "_name_or_path": "",317 "add_cross_attention": false,318 "architectures": null,319 "bad_words_ids": null,320 "begin_suppress_tokens": null,321 "bos_token_id": null,322 "chunk_size_feed_forward": 0,323 "cross_attention_hidden_size": null,324 "decoder_start_token_id": null,325 "depth": 32,326 "diversity_penalty": 0.0,327 "do_sample": false,328 "dtype": "bfloat16",329 "early_stopping": false,330 "embed_dim": 1280,331 "encoder_no_repeat_ngram_size": 0,332 "eos_token_id": null,333 "exponential_decay_length_penalty": null,334 "finetuning_task": null,335 "forced_bos_token_id": null,336 "forced_eos_token_id": null,337 "fullatt_block_indexes": [338 7,339 15,340 23,341 31342 ],343 "hidden_act": "silu",344 "hidden_size": 1280,345 "id2label": {346 "0": "LABEL_0",347 "1": "LABEL_1"348 },349 "in_channels": 3,350 "in_chans": 3,351 "init_std": 0.02,352 "initializer_range": 0.02,353 "intermediate_size": 3420,354 "is_decoder": false,355 "is_encoder_decoder": false,356 "label2id": {357 "LABEL_0": 0,358 "LABEL_1": 1359 },360 "length_penalty": 1.0,361 "max_length": 20,362 "min_length": 0,363 "model_type": "qwen2_5_omni_vision_encoder",364 "no_repeat_ngram_size": 0,365 "num_beam_groups": 1,366 "num_beams": 1,367 "num_heads": 16,368 "num_return_sequences": 1,369 "out_hidden_size": 3584,370 "output_attentions": false,371 "output_hidden_states": false,372 "output_scores": false,373 "pad_token_id": null,374 "patch_size": 14,375 "prefix": null,376 "problem_type": null,377 "pruned_heads": {},378 "remove_invalid_values": false,379 "repetition_penalty": 1.0,380 "return_dict": true,381 "return_dict_in_generate": false,382 "sep_token_id": null,383 "spatial_merge_size": 2,384 "spatial_patch_size": 14,385 "suppress_tokens": null,386 "task_specific_params": null,387 "temperature": 1.0,388 "temporal_patch_size": 2,389 "tf_legacy_loss": false,390 "tie_encoder_decoder": false,391 "tie_word_embeddings": true,392 "tokenizer_class": null,393 "tokens_per_second": 25,394 "top_k": 50,395 "top_p": 1.0,396 "torchscript": false,397 "typical_p": 1.0,398 "use_bfloat16": false,399 "window_size": 112400 },401 "vision_end_token_id": 151653,402 "vision_start_token_id": 151652,403 "vision_token_id": 151654,404 "audio_token_id": 151646,405 "image_token_id": 151655,406 "video_token_id": 151656407 },408 "token2wav_config": {409 "_attn_implementation_autoset": true,410 "bigvgan_config": {411 "_attn_implementation_autoset": true,412 "_name_or_path": "",413 "add_cross_attention": false,414 "architectures": null,415 "bad_words_ids": null,416 "begin_suppress_tokens": null,417 "bos_token_id": null,418 "chunk_size_feed_forward": 0,419 "cross_attention_hidden_size": null,420 "decoder_start_token_id": null,421 "diversity_penalty": 0.0,422 "do_sample": false,423 "dtype": null,424 "early_stopping": false,425 "encoder_no_repeat_ngram_size": 0,426 "eos_token_id": null,427 "exponential_decay_length_penalty": null,428 "finetuning_task": null,429 "forced_bos_token_id": null,430 "forced_eos_token_id": null,431 "id2label": {432 "0": "LABEL_0",433 "1": "LABEL_1"434 },435 "is_decoder": false,436 "is_encoder_decoder": false,437 "label2id": {438 "LABEL_0": 0,439 "LABEL_1": 1440 },441 "length_penalty": 1.0,442 "max_length": 20,443 "mel_dim": 80,444 "min_length": 0,445 "model_type": "qwen2_5_omni_bigvgan",446 "no_repeat_ngram_size": 0,447 "num_beam_groups": 1,448 "num_beams": 1,449 "num_return_sequences": 1,450 "output_attentions": false,451 "output_hidden_states": false,452 "output_scores": false,453 "pad_token_id": null,454 "prefix": null,455 "problem_type": null,456 "pruned_heads": {},457 "remove_invalid_values": false,458 "repetition_penalty": 1.0,459 "resblock_dilation_sizes": [460 [461 1,462 3,463 5464 ],465 [466 1,467 3,468 5469 ],470 [471 1,472 3,473 5474 ]475 ],476 "resblock_kernel_sizes": [477 3,478 7,479 11480 ],481 "return_dict": true,482 "return_dict_in_generate": false,483 "sep_token_id": null,484 "suppress_tokens": null,485 "task_specific_params": null,486 "temperature": 1.0,487 "tf_legacy_loss": false,488 "tie_encoder_decoder": false,489 "tie_word_embeddings": true,490 "tokenizer_class": null,491 "top_k": 50,492 "top_p": 1.0,493 "torchscript": false,494 "typical_p": 1.0,495 "upsample_initial_channel": 1536,496 "upsample_kernel_sizes": [497 11,498 7,499 4,500 4,501 4,502 4503 ],504 "upsample_rates": [505 5,506 3,507 2,508 2,509 2,510 2511 ],512 "use_bfloat16": false,513 "use_bias_at_final": false514 },515 "dit_config": {516 "_attn_implementation_autoset": true,517 "_name_or_path": "",518 "add_cross_attention": false,519 "architectures": null,520 "bad_words_ids": null,521 "begin_suppress_tokens": null,522 "block_size": 24,523 "bos_token_id": null,524 "chunk_size_feed_forward": 0,525 "cross_attention_hidden_size": null,526 "decoder_start_token_id": null,527 "depth": 22,528 "dim": 1024,529 "diversity_penalty": 0.0,530 "do_sample": false,531 "dropout": 0.1,532 "dtype": "float32",533 "early_stopping": false,534 "emb_dim": 512,535 "enc_attention_channels": 64,536 "enc_channels": [537 256,538 256,539 256,540 256,541 768542 ],543 "enc_dilations": [544 1,545 2,546 3,547 4,548 1549 ],550 "enc_dim": 128,551 "enc_emb_dim": 192,552 "enc_global_context": true,553 "enc_kernel_sizes": [554 5,555 3,556 3,557 3,558 1559 ],560 "enc_lin_neurons": 192,561 "enc_res2net_scale": 2,562 "enc_se_channels": 64,563 "encoder_no_repeat_ngram_size": 0,564 "eos_token_id": null,565 "exponential_decay_length_penalty": null,566 "ff_mult": 2,567 "finetuning_task": null,568 "forced_bos_token_id": null,569 "forced_eos_token_id": null,570 "head_dim": 64,571 "heads": 16,572 "hidden_size": 1024,573 "id2label": {574 "0": "LABEL_0",575 "1": "LABEL_1"576 },577 "is_decoder": false,578 "is_encoder_decoder": false,579 "label2id": {580 "LABEL_0": 0,581 "LABEL_1": 1582 },583 "length_penalty": 1.0,584 "look_ahead_layers": [585 10586 ],587 "look_backward_layers": [588 0,589 20590 ],591 "max_length": 20,592 "max_position_embeddings": 32768,593 "mel_dim": 80,594 "min_length": 0,595 "model_type": "qwen2_5_omni_dit",596 "no_repeat_ngram_size": 0,597 "num_attention_heads": 16,598 "num_beam_groups": 1,599 "num_beams": 1,600 "num_embeds": 8193,601 "num_hidden_layers": 22,602 "num_return_sequences": 1,603 "output_attentions": false,604 "output_hidden_states": false,605 "output_scores": false,606 "pad_token_id": null,607 "prefix": null,608 "problem_type": null,609 "pruned_heads": {},610 "remove_invalid_values": false,611 "repeats": 2,612 "repetition_penalty": 1.0,613 "return_dict": true,614 "return_dict_in_generate": false,615 "rope_theta": 10000.0,616 "sep_token_id": null,617 "suppress_tokens": null,618 "task_specific_params": null,619 "temperature": 1.0,620 "tf_legacy_loss": false,621 "tie_encoder_decoder": false,622 "tie_word_embeddings": true,623 "tokenizer_class": null,624 "top_k": 50,625 "top_p": 1.0,626 "torchscript": false,627 "typical_p": 1.0,628 "use_bfloat16": false629 },630 "dtype": "bfloat16",631 "model_type": "qwen2_5_omni_token2wav"632 },633 "transformers_version": "4.57.0.dev0"634}