| { | |
| "architectures": [ | |
| "BailingMMNativeForConditionalGeneration" | |
| ], | |
| "model_type": "bailingmm", | |
| "audio_config": { | |
| "_name_or_path": "", | |
| "add_cross_attention": false, | |
| "architectures": null, | |
| "bad_words_ids": null, | |
| "begin_suppress_tokens": null, | |
| "bos_token_id": null, | |
| "chunk_size_feed_forward": 0, | |
| "cross_attention_hidden_size": null, | |
| "decoder_start_token_id": null, | |
| "diversity_penalty": 0.0, | |
| "do_sample": false, | |
| "ds_kernel_size": 3, | |
| "ds_stride": 2, | |
| "early_stopping": false, | |
| "encoder_no_repeat_ngram_size": 0, | |
| "eos_token_id": null, | |
| "exponential_decay_length_penalty": null, | |
| "finetuning_task": null, | |
| "forced_bos_token_id": null, | |
| "forced_eos_token_id": null, | |
| "id2label": { | |
| "0": "LABEL_0", | |
| "1": "LABEL_1" | |
| }, | |
| "is_decoder": false, | |
| "is_encoder_decoder": false, | |
| "label2id": { | |
| "LABEL_0": 0, | |
| "LABEL_1": 1 | |
| }, | |
| "length_penalty": 1.0, | |
| "max_length": 20, | |
| "min_length": 0, | |
| "model_type": "bailingmm", | |
| "no_repeat_ngram_size": 10, | |
| "norm_query_embeds": true, | |
| "num_beam_groups": 1, | |
| "num_beams": 1, | |
| "num_return_sequences": 1, | |
| "output_attentions": false, | |
| "output_hidden_states": false, | |
| "output_scores": false, | |
| "pad_token_id": null, | |
| "prefix": null, | |
| "problem_type": null, | |
| "pruned_heads": {}, | |
| "remove_invalid_values": false, | |
| "repetition_penalty": 1.0, | |
| "return_dict": true, | |
| "return_dict_in_generate": false, | |
| "sep_token_id": null, | |
| "suppress_tokens": null, | |
| "task_specific_params": null, | |
| "temperature": 1.0, | |
| "tf_legacy_loss": false, | |
| "tie_encoder_decoder": false, | |
| "tie_word_embeddings": true, | |
| "tokenizer_class": null, | |
| "top_k": 50, | |
| "top_p": 1.0, | |
| "torch_dtype": null, | |
| "torchscript": false, | |
| "typical_p": 1.0, | |
| "use_bfloat16": false, | |
| "whisper_encoder_config": { | |
| "n_ctx": 15000, | |
| "n_head": 20, | |
| "n_layer": 32, | |
| "n_mels": 128, | |
| "n_state": 1280 | |
| } | |
| }, | |
| "auto_map": { | |
| "AutoConfig": "configuration_bailingmm.BailingMMConfig" | |
| }, | |
| "llm_config": { | |
| "_attn_implementation_autoset": true, | |
| "_name_or_path": "", | |
| "add_cross_attention": false, | |
| "architectures": [ | |
| "BailingMoeForCausalLM" | |
| ], | |
| "attention_dropout": 0.0, | |
| "audio_end_token": 126359, | |
| "audio_patch_token": 126357, | |
| "audio_start_token": 126358, | |
| "auto_map": { | |
| "AutoConfig": "configuration_bailing_moe.BailingMoeConfig", | |
| "AutoModel": "modeling_bailing_moe.BailingMoeModel", | |
| "AutoModelForCausalLM": "modeling_bailing_moe.BailingMoeForCausalLM", | |
| "AutoModelForTokenClassification": "modeling_bailing_moe.BailingMoeForTokenClassification" | |
| }, | |
| "bad_words_ids": null, | |
| "begin_suppress_tokens": null, | |
| "bos_token_id": null, | |
| "chunk_size_feed_forward": 0, | |
| "cross_attention_hidden_size": null, | |
| "decoder_start_token_id": null, | |
| "diversity_penalty": 0.0, | |
| "do_sample": false, | |
| "early_stopping": false, | |
| "embedding_dropout": 0.0, | |
| "encoder_no_repeat_ngram_size": 0, | |
| "eos_token_id": 126081, | |
| "exponential_decay_length_penalty": null, | |
| "finetuning_task": null, | |
| "first_k_dense_replace": 0, | |
| "forced_bos_token_id": null, | |
| "forced_eos_token_id": null, | |
| "head_dim": 128, | |
| "hidden_act": "silu", | |
| "hidden_size": 2048, | |
| "id2label": { | |
| "0": "LABEL_0", | |
| "1": "LABEL_1" | |
| }, | |
| "image_end_token": 126348, | |
| "image_patch_token": 126346, | |
| "image_start_token": 126347, | |
| "initializer_range": 0.006, | |
| "intermediate_size": 5632, | |
| "is_decoder": false, | |
| "is_encoder_decoder": false, | |
| "label2id": { | |
| "LABEL_0": 0, | |
| "LABEL_1": 1 | |
| }, | |
| "length_penalty": 1.0, | |
| "max_length": 20, | |
| "max_position_embeddings": 32768, | |
| "max_window_layers": 28, | |
| "min_length": 0, | |
| "model_type": "bailing_moe", | |
| "moe_intermediate_size": 1408, | |
| "multi_gate": true, | |
| "no_repeat_ngram_size": 0, | |
| "norm_head": false, | |
| "norm_softmax": false, | |
| "norm_topk_prob": true, | |
| "num_attention_heads": 16, | |
| "num_beam_groups": 1, | |
| "num_beams": 1, | |
| "num_experts": 64, | |
| "num_experts_per_tok": 6, | |
| "num_hidden_layers": 28, | |
| "num_key_value_heads": 4, | |
| "num_return_sequences": 1, | |
| "num_shared_experts": 2, | |
| "output_attentions": false, | |
| "output_dropout": 0.0, | |
| "output_hidden_states": false, | |
| "output_router_logits": false, | |
| "output_scores": false, | |
| "pad_token_id": 126081, | |
| "prefix": null, | |
| "pretraining_tp": 1, | |
| "problem_type": null, | |
| "pruned_heads": {}, | |
| "remove_invalid_values": false, | |
| "repetition_penalty": 1.0, | |
| "return_dict": true, | |
| "return_dict_in_generate": false, | |
| "rms_norm_eps": 1e-05, | |
| "rope_scaling": { | |
| "type": "3D", | |
| "rope_type": "default", | |
| "mrope_section": [ | |
| 16, | |
| 24, | |
| 24 | |
| ], | |
| "use_3d_rope": true | |
| }, | |
| "rope_theta": 600000, | |
| "sep_token_id": null, | |
| "sliding_window": 4096, | |
| "suppress_tokens": null, | |
| "task_specific_params": null, | |
| "temperature": 1.0, | |
| "tf_legacy_loss": false, | |
| "tie_encoder_decoder": false, | |
| "tie_word_embeddings": false, | |
| "tokenizer_class": null, | |
| "top_k": 50, | |
| "top_p": 1.0, | |
| "torch_dtype": "bfloat16", | |
| "torchscript": false, | |
| "typical_p": 1.0, | |
| "use_bfloat16": false, | |
| "use_bias": false, | |
| "use_cache": true, | |
| "use_qkv_bias": false, | |
| "use_sliding_window": false, | |
| "video_end_token": 126350, | |
| "video_start_token": 126349, | |
| "vocab_size": 126464 | |
| }, | |
| "mlp_depth": 2, | |
| "model_type": "bailingmm", | |
| "quantization_config": { | |
| "config_groups": { | |
| "group_0": { | |
| "input_activations": { | |
| "actorder": null, | |
| "block_structure": null, | |
| "dynamic": true, | |
| "group_size": null, | |
| "num_bits": 8, | |
| "observer": null, | |
| "observer_kwargs": {}, | |
| "strategy": "token", | |
| "symmetric": true, | |
| "type": "float" | |
| }, | |
| "output_activations": null, | |
| "targets": [ | |
| "Linear" | |
| ], | |
| "weights": { | |
| "actorder": null, | |
| "block_structure": null, | |
| "dynamic": false, | |
| "group_size": null, | |
| "num_bits": 8, | |
| "observer": "minmax", | |
| "observer_kwargs": {}, | |
| "strategy": "channel", | |
| "symmetric": true, | |
| "type": "float" | |
| } | |
| } | |
| }, | |
| "format": "float-quantized", | |
| "global_compression_ratio": 1.5072876165969304, | |
| "ignore": [ | |
| "vision.blocks.0.attn.qkv", | |
| "vision.blocks.0.attn.proj", | |
| "vision.blocks.0.mlp.gate_proj", | |
| "vision.blocks.0.mlp.up_proj", | |
| "vision.blocks.0.mlp.down_proj", | |
| "vision.blocks.1.attn.qkv", | |
| "vision.blocks.1.attn.proj", | |
| "vision.blocks.1.mlp.gate_proj", | |
| "vision.blocks.1.mlp.up_proj", | |
| "vision.blocks.1.mlp.down_proj", | |
| "vision.blocks.2.attn.qkv", | |
| "vision.blocks.2.attn.proj", | |
| "vision.blocks.2.mlp.gate_proj", | |
| "vision.blocks.2.mlp.up_proj", | |
| "vision.blocks.2.mlp.down_proj", | |
| "vision.blocks.3.attn.qkv", | |
| "vision.blocks.3.attn.proj", | |
| "vision.blocks.3.mlp.gate_proj", | |
| "vision.blocks.3.mlp.up_proj", | |
| "vision.blocks.3.mlp.down_proj", | |
| "vision.blocks.4.attn.qkv", | |
| "vision.blocks.4.attn.proj", | |
| "vision.blocks.4.mlp.gate_proj", | |
| "vision.blocks.4.mlp.up_proj", | |
| "vision.blocks.4.mlp.down_proj", | |
| "vision.blocks.5.attn.qkv", | |
| "vision.blocks.5.attn.proj", | |
| "vision.blocks.5.mlp.gate_proj", | |
| "vision.blocks.5.mlp.up_proj", | |
| "vision.blocks.5.mlp.down_proj", | |
| "vision.blocks.6.attn.qkv", | |
| "vision.blocks.6.attn.proj", | |
| "vision.blocks.6.mlp.gate_proj", | |
| "vision.blocks.6.mlp.up_proj", | |
| "vision.blocks.6.mlp.down_proj", | |
| "vision.blocks.7.attn.qkv", | |
| "vision.blocks.7.attn.proj", | |
| "vision.blocks.7.mlp.gate_proj", | |
| "vision.blocks.7.mlp.up_proj", | |
| "vision.blocks.7.mlp.down_proj", | |
| "vision.blocks.8.attn.qkv", | |
| "vision.blocks.8.attn.proj", | |
| "vision.blocks.8.mlp.gate_proj", | |
| "vision.blocks.8.mlp.up_proj", | |
| "vision.blocks.8.mlp.down_proj", | |
| "vision.blocks.9.attn.qkv", | |
| "vision.blocks.9.attn.proj", | |
| "vision.blocks.9.mlp.gate_proj", | |
| "vision.blocks.9.mlp.up_proj", | |
| "vision.blocks.9.mlp.down_proj", | |
| "vision.blocks.10.attn.qkv", | |
| "vision.blocks.10.attn.proj", | |
| "vision.blocks.10.mlp.gate_proj", | |
| "vision.blocks.10.mlp.up_proj", | |
| "vision.blocks.10.mlp.down_proj", | |
| "vision.blocks.11.attn.qkv", | |
| "vision.blocks.11.attn.proj", | |
| "vision.blocks.11.mlp.gate_proj", | |
| "vision.blocks.11.mlp.up_proj", | |
| "vision.blocks.11.mlp.down_proj", | |
| "vision.blocks.12.attn.qkv", | |
| "vision.blocks.12.attn.proj", | |
| "vision.blocks.12.mlp.gate_proj", | |
| "vision.blocks.12.mlp.up_proj", | |
| "vision.blocks.12.mlp.down_proj", | |
| "vision.blocks.13.attn.qkv", | |
| "vision.blocks.13.attn.proj", | |
| "vision.blocks.13.mlp.gate_proj", | |
| "vision.blocks.13.mlp.up_proj", | |
| "vision.blocks.13.mlp.down_proj", | |
| "vision.blocks.14.attn.qkv", | |
| "vision.blocks.14.attn.proj", | |
| "vision.blocks.14.mlp.gate_proj", | |
| "vision.blocks.14.mlp.up_proj", | |
| "vision.blocks.14.mlp.down_proj", | |
| "vision.blocks.15.attn.qkv", | |
| "vision.blocks.15.attn.proj", | |
| "vision.blocks.15.mlp.gate_proj", | |
| "vision.blocks.15.mlp.up_proj", | |
| "vision.blocks.15.mlp.down_proj", | |
| "vision.blocks.16.attn.qkv", | |
| "vision.blocks.16.attn.proj", | |
| "vision.blocks.16.mlp.gate_proj", | |
| "vision.blocks.16.mlp.up_proj", | |
| "vision.blocks.16.mlp.down_proj", | |
| "vision.blocks.17.attn.qkv", | |
| "vision.blocks.17.attn.proj", | |
| "vision.blocks.17.mlp.gate_proj", | |
| "vision.blocks.17.mlp.up_proj", | |
| "vision.blocks.17.mlp.down_proj", | |
| "vision.blocks.18.attn.qkv", | |
| "vision.blocks.18.attn.proj", | |
| "vision.blocks.18.mlp.gate_proj", | |
| "vision.blocks.18.mlp.up_proj", | |
| "vision.blocks.18.mlp.down_proj", | |
| "vision.blocks.19.attn.qkv", | |
| "vision.blocks.19.attn.proj", | |
| "vision.blocks.19.mlp.gate_proj", | |
| "vision.blocks.19.mlp.up_proj", | |
| "vision.blocks.19.mlp.down_proj", | |
| "vision.blocks.20.attn.qkv", | |
| "vision.blocks.20.attn.proj", | |
| "vision.blocks.20.mlp.gate_proj", | |
| "vision.blocks.20.mlp.up_proj", | |
| "vision.blocks.20.mlp.down_proj", | |
| "vision.blocks.21.attn.qkv", | |
| "vision.blocks.21.attn.proj", | |
| "vision.blocks.21.mlp.gate_proj", | |
| "vision.blocks.21.mlp.up_proj", | |
| "vision.blocks.21.mlp.down_proj", | |
| "vision.blocks.22.attn.qkv", | |
| "vision.blocks.22.attn.proj", | |
| "vision.blocks.22.mlp.gate_proj", | |
| "vision.blocks.22.mlp.up_proj", | |
| "vision.blocks.22.mlp.down_proj", | |
| "vision.blocks.23.attn.qkv", | |
| "vision.blocks.23.attn.proj", | |
| "vision.blocks.23.mlp.gate_proj", | |
| "vision.blocks.23.mlp.up_proj", | |
| "vision.blocks.23.mlp.down_proj", | |
| "vision.blocks.24.attn.qkv", | |
| "vision.blocks.24.attn.proj", | |
| "vision.blocks.24.mlp.gate_proj", | |
| "vision.blocks.24.mlp.up_proj", | |
| "vision.blocks.24.mlp.down_proj", | |
| "vision.blocks.25.attn.qkv", | |
| "vision.blocks.25.attn.proj", | |
| "vision.blocks.25.mlp.gate_proj", | |
| "vision.blocks.25.mlp.up_proj", | |
| "vision.blocks.25.mlp.down_proj", | |
| "vision.blocks.26.attn.qkv", | |
| "vision.blocks.26.attn.proj", | |
| "vision.blocks.26.mlp.gate_proj", | |
| "vision.blocks.26.mlp.up_proj", | |
| "vision.blocks.26.mlp.down_proj", | |
| "vision.blocks.27.attn.qkv", | |
| "vision.blocks.27.attn.proj", | |
| "vision.blocks.27.mlp.gate_proj", | |
| "vision.blocks.27.mlp.up_proj", | |
| "vision.blocks.27.mlp.down_proj", | |
| "vision.blocks.28.attn.qkv", | |
| "vision.blocks.28.attn.proj", | |
| "vision.blocks.28.mlp.gate_proj", | |
| "vision.blocks.28.mlp.up_proj", | |
| "vision.blocks.28.mlp.down_proj", | |
| "vision.blocks.29.attn.qkv", | |
| "vision.blocks.29.attn.proj", | |
| "vision.blocks.29.mlp.gate_proj", | |
| "vision.blocks.29.mlp.up_proj", | |
| "vision.blocks.29.mlp.down_proj", | |
| "vision.blocks.30.attn.qkv", | |
| "vision.blocks.30.attn.proj", | |
| "vision.blocks.30.mlp.gate_proj", | |
| "vision.blocks.30.mlp.up_proj", | |
| "vision.blocks.30.mlp.down_proj", | |
| "vision.blocks.31.attn.qkv", | |
| "vision.blocks.31.attn.proj", | |
| "vision.blocks.31.mlp.gate_proj", | |
| "vision.blocks.31.mlp.up_proj", | |
| "vision.blocks.31.mlp.down_proj", | |
| "vision.merger.mlp.0", | |
| "vision.merger.mlp.2", | |
| "audio.blocks.0.attn.query", | |
| "audio.blocks.0.attn.key", | |
| "audio.blocks.0.attn.value", | |
| "audio.blocks.0.attn.out", | |
| "audio.blocks.0.mlp.0", | |
| "audio.blocks.0.mlp.2", | |
| "audio.blocks.1.attn.query", | |
| "audio.blocks.1.attn.key", | |
| "audio.blocks.1.attn.value", | |
| "audio.blocks.1.attn.out", | |
| "audio.blocks.1.mlp.0", | |
| "audio.blocks.1.mlp.2", | |
| "audio.blocks.2.attn.query", | |
| "audio.blocks.2.attn.key", | |
| "audio.blocks.2.attn.value", | |
| "audio.blocks.2.attn.out", | |
| "audio.blocks.2.mlp.0", | |
| "audio.blocks.2.mlp.2", | |
| "audio.blocks.3.attn.query", | |
| "audio.blocks.3.attn.key", | |
| "audio.blocks.3.attn.value", | |
| "audio.blocks.3.attn.out", | |
| "audio.blocks.3.mlp.0", | |
| "audio.blocks.3.mlp.2", | |
| "audio.blocks.4.attn.query", | |
| "audio.blocks.4.attn.key", | |
| "audio.blocks.4.attn.value", | |
| "audio.blocks.4.attn.out", | |
| "audio.blocks.4.mlp.0", | |
| "audio.blocks.4.mlp.2", | |
| "audio.blocks.5.attn.query", | |
| "audio.blocks.5.attn.key", | |
| "audio.blocks.5.attn.value", | |
| "audio.blocks.5.attn.out", | |
| "audio.blocks.5.mlp.0", | |
| "audio.blocks.5.mlp.2", | |
| "audio.blocks.6.attn.query", | |
| "audio.blocks.6.attn.key", | |
| "audio.blocks.6.attn.value", | |
| "audio.blocks.6.attn.out", | |
| "audio.blocks.6.mlp.0", | |
| "audio.blocks.6.mlp.2", | |
| "audio.blocks.7.attn.query", | |
| "audio.blocks.7.attn.key", | |
| "audio.blocks.7.attn.value", | |
| "audio.blocks.7.attn.out", | |
| "audio.blocks.7.mlp.0", | |
| "audio.blocks.7.mlp.2", | |
| "audio.blocks.8.attn.query", | |
| "audio.blocks.8.attn.key", | |
| "audio.blocks.8.attn.value", | |
| "audio.blocks.8.attn.out", | |
| "audio.blocks.8.mlp.0", | |
| "audio.blocks.8.mlp.2", | |
| "audio.blocks.9.attn.query", | |
| "audio.blocks.9.attn.key", | |
| "audio.blocks.9.attn.value", | |
| "audio.blocks.9.attn.out", | |
| "audio.blocks.9.mlp.0", | |
| "audio.blocks.9.mlp.2", | |
| "audio.blocks.10.attn.query", | |
| "audio.blocks.10.attn.key", | |
| "audio.blocks.10.attn.value", | |
| "audio.blocks.10.attn.out", | |
| "audio.blocks.10.mlp.0", | |
| "audio.blocks.10.mlp.2", | |
| "audio.blocks.11.attn.query", | |
| "audio.blocks.11.attn.key", | |
| "audio.blocks.11.attn.value", | |
| "audio.blocks.11.attn.out", | |
| "audio.blocks.11.mlp.0", | |
| "audio.blocks.11.mlp.2", | |
| "audio.blocks.12.attn.query", | |
| "audio.blocks.12.attn.key", | |
| "audio.blocks.12.attn.value", | |
| "audio.blocks.12.attn.out", | |
| "audio.blocks.12.mlp.0", | |
| "audio.blocks.12.mlp.2", | |
| "audio.blocks.13.attn.query", | |
| "audio.blocks.13.attn.key", | |
| "audio.blocks.13.attn.value", | |
| "audio.blocks.13.attn.out", | |
| "audio.blocks.13.mlp.0", | |
| "audio.blocks.13.mlp.2", | |
| "audio.blocks.14.attn.query", | |
| "audio.blocks.14.attn.key", | |
| "audio.blocks.14.attn.value", | |
| "audio.blocks.14.attn.out", | |
| "audio.blocks.14.mlp.0", | |
| "audio.blocks.14.mlp.2", | |
| "audio.blocks.15.attn.query", | |
| "audio.blocks.15.attn.key", | |
| "audio.blocks.15.attn.value", | |
| "audio.blocks.15.attn.out", | |
| "audio.blocks.15.mlp.0", | |
| "audio.blocks.15.mlp.2", | |
| "audio.blocks.16.attn.query", | |
| "audio.blocks.16.attn.key", | |
| "audio.blocks.16.attn.value", | |
| "audio.blocks.16.attn.out", | |
| "audio.blocks.16.mlp.0", | |
| "audio.blocks.16.mlp.2", | |
| "audio.blocks.17.attn.query", | |
| "audio.blocks.17.attn.key", | |
| "audio.blocks.17.attn.value", | |
| "audio.blocks.17.attn.out", | |
| "audio.blocks.17.mlp.0", | |
| "audio.blocks.17.mlp.2", | |
| "audio.blocks.18.attn.query", | |
| "audio.blocks.18.attn.key", | |
| "audio.blocks.18.attn.value", | |
| "audio.blocks.18.attn.out", | |
| "audio.blocks.18.mlp.0", | |
| "audio.blocks.18.mlp.2", | |
| "audio.blocks.19.attn.query", | |
| "audio.blocks.19.attn.key", | |
| "audio.blocks.19.attn.value", | |
| "audio.blocks.19.attn.out", | |
| "audio.blocks.19.mlp.0", | |
| "audio.blocks.19.mlp.2", | |
| "audio.blocks.20.attn.query", | |
| "audio.blocks.20.attn.key", | |
| "audio.blocks.20.attn.value", | |
| "audio.blocks.20.attn.out", | |
| "audio.blocks.20.mlp.0", | |
| "audio.blocks.20.mlp.2", | |
| "audio.blocks.21.attn.query", | |
| "audio.blocks.21.attn.key", | |
| "audio.blocks.21.attn.value", | |
| "audio.blocks.21.attn.out", | |
| "audio.blocks.21.mlp.0", | |
| "audio.blocks.21.mlp.2", | |
| "audio.blocks.22.attn.query", | |
| "audio.blocks.22.attn.key", | |
| "audio.blocks.22.attn.value", | |
| "audio.blocks.22.attn.out", | |
| "audio.blocks.22.mlp.0", | |
| "audio.blocks.22.mlp.2", | |
| "audio.blocks.23.attn.query", | |
| "audio.blocks.23.attn.key", | |
| "audio.blocks.23.attn.value", | |
| "audio.blocks.23.attn.out", | |
| "audio.blocks.23.mlp.0", | |
| "audio.blocks.23.mlp.2", | |
| "audio.blocks.24.attn.query", | |
| "audio.blocks.24.attn.key", | |
| "audio.blocks.24.attn.value", | |
| "audio.blocks.24.attn.out", | |
| "audio.blocks.24.mlp.0", | |
| "audio.blocks.24.mlp.2", | |
| "audio.blocks.25.attn.query", | |
| "audio.blocks.25.attn.key", | |
| "audio.blocks.25.attn.value", | |
| "audio.blocks.25.attn.out", | |
| "audio.blocks.25.mlp.0", | |
| "audio.blocks.25.mlp.2", | |
| "audio.blocks.26.attn.query", | |
| "audio.blocks.26.attn.key", | |
| "audio.blocks.26.attn.value", | |
| "audio.blocks.26.attn.out", | |
| "audio.blocks.26.mlp.0", | |
| "audio.blocks.26.mlp.2", | |
| "audio.blocks.27.attn.query", | |
| "audio.blocks.27.attn.key", | |
| "audio.blocks.27.attn.value", | |
| "audio.blocks.27.attn.out", | |
| "audio.blocks.27.mlp.0", | |
| "audio.blocks.27.mlp.2", | |
| "audio.blocks.28.attn.query", | |
| "audio.blocks.28.attn.key", | |
| "audio.blocks.28.attn.value", | |
| "audio.blocks.28.attn.out", | |
| "audio.blocks.28.mlp.0", | |
| "audio.blocks.28.mlp.2", | |
| "audio.blocks.29.attn.query", | |
| "audio.blocks.29.attn.key", | |
| "audio.blocks.29.attn.value", | |
| "audio.blocks.29.attn.out", | |
| "audio.blocks.29.mlp.0", | |
| "audio.blocks.29.mlp.2", | |
| "audio.blocks.30.attn.query", | |
| "audio.blocks.30.attn.key", | |
| "audio.blocks.30.attn.value", | |
| "audio.blocks.30.attn.out", | |
| "audio.blocks.30.mlp.0", | |
| "audio.blocks.30.mlp.2", | |
| "audio.blocks.31.attn.query", | |
| "audio.blocks.31.attn.key", | |
| "audio.blocks.31.attn.value", | |
| "audio.blocks.31.attn.out", | |
| "audio.blocks.31.mlp.0", | |
| "audio.blocks.31.mlp.2", | |
| "model.model.layers.0.mlp.shared_experts.gate_proj", | |
| "model.model.layers.0.mlp.shared_experts.up_proj", | |
| "model.model.layers.0.mlp.shared_experts.down_proj", | |
| "model.model.layers.1.mlp.shared_experts.gate_proj", | |
| "model.model.layers.1.mlp.shared_experts.up_proj", | |
| "model.model.layers.1.mlp.shared_experts.down_proj", | |
| "model.model.layers.2.mlp.shared_experts.gate_proj", | |
| "model.model.layers.2.mlp.shared_experts.up_proj", | |
| "model.model.layers.2.mlp.shared_experts.down_proj", | |
| "model.model.layers.3.mlp.shared_experts.gate_proj", | |
| "model.model.layers.3.mlp.shared_experts.up_proj", | |
| "model.model.layers.3.mlp.shared_experts.down_proj", | |
| "model.model.layers.4.mlp.shared_experts.gate_proj", | |
| "model.model.layers.4.mlp.shared_experts.up_proj", | |
| "model.model.layers.4.mlp.shared_experts.down_proj", | |
| "model.model.layers.5.mlp.shared_experts.gate_proj", | |
| "model.model.layers.5.mlp.shared_experts.up_proj", | |
| "model.model.layers.5.mlp.shared_experts.down_proj", | |
| "model.model.layers.6.mlp.shared_experts.gate_proj", | |
| "model.model.layers.6.mlp.shared_experts.up_proj", | |
| "model.model.layers.6.mlp.shared_experts.down_proj", | |
| "model.model.layers.7.mlp.shared_experts.gate_proj", | |
| "model.model.layers.7.mlp.shared_experts.up_proj", | |
| "model.model.layers.7.mlp.shared_experts.down_proj", | |
| "model.model.layers.8.mlp.shared_experts.gate_proj", | |
| "model.model.layers.8.mlp.shared_experts.up_proj", | |
| "model.model.layers.8.mlp.shared_experts.down_proj", | |
| "model.model.layers.9.mlp.shared_experts.gate_proj", | |
| "model.model.layers.9.mlp.shared_experts.up_proj", | |
| "model.model.layers.9.mlp.shared_experts.down_proj", | |
| "model.model.layers.10.mlp.shared_experts.gate_proj", | |
| "model.model.layers.10.mlp.shared_experts.up_proj", | |
| "model.model.layers.10.mlp.shared_experts.down_proj", | |
| "model.model.layers.11.mlp.shared_experts.gate_proj", | |
| "model.model.layers.11.mlp.shared_experts.up_proj", | |
| "model.model.layers.11.mlp.shared_experts.down_proj", | |
| "model.model.layers.12.mlp.shared_experts.gate_proj", | |
| "model.model.layers.12.mlp.shared_experts.up_proj", | |
| "model.model.layers.12.mlp.shared_experts.down_proj", | |
| "model.model.layers.13.mlp.shared_experts.gate_proj", | |
| "model.model.layers.13.mlp.shared_experts.up_proj", | |
| "model.model.layers.13.mlp.shared_experts.down_proj", | |
| "model.model.layers.14.mlp.shared_experts.gate_proj", | |
| "model.model.layers.14.mlp.shared_experts.up_proj", | |
| "model.model.layers.14.mlp.shared_experts.down_proj", | |
| "model.model.layers.15.mlp.shared_experts.gate_proj", | |
| "model.model.layers.15.mlp.shared_experts.up_proj", | |
| "model.model.layers.15.mlp.shared_experts.down_proj", | |
| "model.model.layers.16.mlp.shared_experts.gate_proj", | |
| "model.model.layers.16.mlp.shared_experts.up_proj", | |
| "model.model.layers.16.mlp.shared_experts.down_proj", | |
| "model.model.layers.17.mlp.shared_experts.gate_proj", | |
| "model.model.layers.17.mlp.shared_experts.up_proj", | |
| "model.model.layers.17.mlp.shared_experts.down_proj", | |
| "model.model.layers.18.mlp.shared_experts.gate_proj", | |
| "model.model.layers.18.mlp.shared_experts.up_proj", | |
| "model.model.layers.18.mlp.shared_experts.down_proj", | |
| "model.model.layers.19.mlp.shared_experts.gate_proj", | |
| "model.model.layers.19.mlp.shared_experts.up_proj", | |
| "model.model.layers.19.mlp.shared_experts.down_proj", | |
| "model.model.layers.20.mlp.shared_experts.gate_proj", | |
| "model.model.layers.20.mlp.shared_experts.up_proj", | |
| "model.model.layers.20.mlp.shared_experts.down_proj", | |
| "model.model.layers.21.mlp.shared_experts.gate_proj", | |
| "model.model.layers.21.mlp.shared_experts.up_proj", | |
| "model.model.layers.21.mlp.shared_experts.down_proj", | |
| "model.model.layers.22.mlp.shared_experts.gate_proj", | |
| "model.model.layers.22.mlp.shared_experts.up_proj", | |
| "model.model.layers.22.mlp.shared_experts.down_proj", | |
| "model.model.layers.23.mlp.shared_experts.gate_proj", | |
| "model.model.layers.23.mlp.shared_experts.up_proj", | |
| "model.model.layers.23.mlp.shared_experts.down_proj", | |
| "model.model.layers.24.mlp.shared_experts.gate_proj", | |
| "model.model.layers.24.mlp.shared_experts.up_proj", | |
| "model.model.layers.24.mlp.shared_experts.down_proj", | |
| "model.model.layers.25.mlp.shared_experts.gate_proj", | |
| "model.model.layers.25.mlp.shared_experts.up_proj", | |
| "model.model.layers.25.mlp.shared_experts.down_proj", | |
| "model.model.layers.26.mlp.shared_experts.gate_proj", | |
| "model.model.layers.26.mlp.shared_experts.up_proj", | |
| "model.model.layers.26.mlp.shared_experts.down_proj", | |
| "model.model.layers.27.mlp.shared_experts.gate_proj", | |
| "model.model.layers.27.mlp.shared_experts.up_proj", | |
| "model.model.layers.27.mlp.shared_experts.down_proj", | |
| "model.lm_head", | |
| "linear_proj.0", | |
| "linear_proj.2", | |
| "linear_proj_audio.3" | |
| ], | |
| "kv_cache_scheme": null, | |
| "quant_method": "compressed-tensors", | |
| "quantization_status": "compressed" | |
| }, | |
| "talker_config": null, | |
| "torch_dtype": "bfloat16", | |
| "transformers_version": "4.52.4", | |
| "vision_config": { | |
| "_attn_implementation_autoset": true, | |
| "_name_or_path": "", | |
| "add_cross_attention": false, | |
| "architectures": [ | |
| "Qwen2_5_VisionTransformer" | |
| ], | |
| "auto_map": { | |
| "AutoConfig": "configuration_qwen2_5_vit.Qwen2_5_VLVisionConfig", | |
| "AutoModel": "qwen2_5_vit.Qwen2_5_VisionTransformer" | |
| }, | |
| "bad_words_ids": null, | |
| "begin_suppress_tokens": null, | |
| "bos_token_id": null, | |
| "chunk_size_feed_forward": 0, | |
| "cross_attention_hidden_size": null, | |
| "decoder_start_token_id": null, | |
| "depth": 32, | |
| "diversity_penalty": 0.0, | |
| "do_sample": false, | |
| "early_stopping": false, | |
| "encoder_no_repeat_ngram_size": 0, | |
| "eos_token_id": null, | |
| "exponential_decay_length_penalty": null, | |
| "finetuning_task": null, | |
| "forced_bos_token_id": null, | |
| "forced_eos_token_id": null, | |
| "fullatt_block_indexes": [ | |
| 7, | |
| 15, | |
| 23, | |
| 31 | |
| ], | |
| "hidden_act": "silu", | |
| "hidden_size": 1280, | |
| "id2label": { | |
| "0": "LABEL_0", | |
| "1": "LABEL_1" | |
| }, | |
| "in_channels": 3, | |
| "in_chans": 3, | |
| "intermediate_size": 3456, | |
| "is_decoder": false, | |
| "is_encoder_decoder": false, | |
| "label2id": { | |
| "LABEL_0": 0, | |
| "LABEL_1": 1 | |
| }, | |
| "length_penalty": 1.0, | |
| "max_length": 20, | |
| "min_length": 0, | |
| "model_type": "qwen2_5_vit", | |
| "no_repeat_ngram_size": 0, | |
| "num_beam_groups": 1, | |
| "num_beams": 1, | |
| "num_heads": 16, | |
| "num_return_sequences": 1, | |
| "out_hidden_size": 8192, | |
| "output_attentions": false, | |
| "output_hidden_states": false, | |
| "output_scores": false, | |
| "pad_token_id": null, | |
| "patch_size": 14, | |
| "prefix": null, | |
| "problem_type": null, | |
| "pruned_heads": {}, | |
| "remove_invalid_values": false, | |
| "repetition_penalty": 1.0, | |
| "return_dict": true, | |
| "return_dict_in_generate": false, | |
| "sep_token_id": null, | |
| "spatial_merge_size": 2, | |
| "spatial_patch_size": 14, | |
| "suppress_tokens": null, | |
| "task_specific_params": null, | |
| "temperature": 1.0, | |
| "temporal_patch_size": 2, | |
| "tf_legacy_loss": false, | |
| "tie_encoder_decoder": false, | |
| "tie_word_embeddings": true, | |
| "tokenizer_class": null, | |
| "tokens_per_second": 2, | |
| "top_k": 50, | |
| "top_p": 1.0, | |
| "torch_dtype": "bfloat16", | |
| "torchscript": false, | |
| "typical_p": 1.0, | |
| "use_bfloat16": false, | |
| "window_size": 112 | |
| } | |
| } | |