| { | |
| "add_classification_head": false, | |
| "anyres_pooling_size": 2, | |
| "anyres_vit_max_image_size": null, | |
| "anyres_vit_two_views": false, | |
| "architectures": [ | |
| "HunyuanImage3ForCausalMM" | |
| ], | |
| "attention_bias": false, | |
| "attention_dropout": 0.0, | |
| "attention_head_dim": 128, | |
| "auto_map": { | |
| "AutoConfig": "configuration_hunyuan_image_3.HunyuanImage3Config", | |
| "AutoModel": "modeling_hunyuan_image_3.HunyuanImage3Model", | |
| "AutoModelForCausalLM": "modeling_hunyuan_image_3.HunyuanImage3ForCausalMM" | |
| }, | |
| "bos_token_id": 127958, | |
| "capacity_factor": 1.0, | |
| "cfg_distilled": true, | |
| "cla_share_factor": 2, | |
| "class_num": 0, | |
| "cond_image_type": "vae_vit", | |
| "cond_token_attn_type": "joint_full", | |
| "dense_list": [ | |
| 4096, | |
| 0 | |
| ], | |
| "dtype": "bfloat16", | |
| "eod_token_id": 3, | |
| "eos_token_id": 127957, | |
| "group_limited_greedy": false, | |
| "head_dim": 128, | |
| "hidden_act": "silu", | |
| "hidden_size": 4096, | |
| "im_end_id": 128001, | |
| "im_newline_id": 11, | |
| "im_start_id": 128000, | |
| "image_base_size": 1024, | |
| "image_token_id": 128006, | |
| "img_proj_type": "unet", | |
| "initializer_range": 0.02, | |
| "intermediate_size": 3072, | |
| "kv_lora_rank": null, | |
| "mask_init_id": 12, | |
| "max_position_embeddings": 22800, | |
| "mlp_bias": false, | |
| "model_type": "Hunyuan", | |
| "moe_drop_tokens": false, | |
| "moe_impl": "eager", | |
| "moe_intermediate_size": [ | |
| 3072, | |
| 3072, | |
| 3072, | |
| 3072, | |
| 3072, | |
| 3072, | |
| 3072, | |
| 3072, | |
| 3072, | |
| 3072, | |
| 3072, | |
| 3072, | |
| 3072, | |
| 3072, | |
| 3072, | |
| 3072, | |
| 3072, | |
| 3072, | |
| 3072, | |
| 3072, | |
| 3072, | |
| 3072, | |
| 3072, | |
| 3072, | |
| 3072, | |
| 3072, | |
| 3072, | |
| 3072, | |
| 3072, | |
| 3072, | |
| 3072, | |
| 3072 | |
| ], | |
| "moe_layer_num_skipped": 0, | |
| "moe_random_routing_dropped_token": false, | |
| "moe_topk": [ | |
| 8, | |
| 8, | |
| 8, | |
| 8, | |
| 8, | |
| 8, | |
| 8, | |
| 8, | |
| 8, | |
| 8, | |
| 8, | |
| 8, | |
| 8, | |
| 8, | |
| 8, | |
| 8, | |
| 8, | |
| 8, | |
| 8, | |
| 8, | |
| 8, | |
| 8, | |
| 8, | |
| 8, | |
| 8, | |
| 8, | |
| 8, | |
| 8, | |
| 8, | |
| 8, | |
| 8, | |
| 8 | |
| ], | |
| "n_group": false, | |
| "norm_topk_prob": true, | |
| "norm_type": "rms", | |
| "num_attention_heads": 32, | |
| "num_experts": 64, | |
| "num_hidden_layers": 32, | |
| "num_key_value_heads": 8, | |
| "num_media_embeds": 257, | |
| "num_shared_expert": [ | |
| 1, | |
| 1, | |
| 1, | |
| 1, | |
| 1, | |
| 1, | |
| 1, | |
| 1, | |
| 1, | |
| 1, | |
| 1, | |
| 1, | |
| 1, | |
| 1, | |
| 1, | |
| 1, | |
| 1, | |
| 1, | |
| 1, | |
| 1, | |
| 1, | |
| 1, | |
| 1, | |
| 1, | |
| 1, | |
| 1, | |
| 1, | |
| 1, | |
| 1, | |
| 1, | |
| 1, | |
| 1 | |
| ], | |
| "pad_id": 128009, | |
| "pad_token_id": 128009, | |
| "patch_embed_hidden_dim": 1024, | |
| "patch_size": 1, | |
| "pool_type": "last", | |
| "position_embedding_xdrope": false, | |
| "pretraining_tp": 1, | |
| "q_lora_rank": null, | |
| "qk_nope_head_dim": null, | |
| "qk_rope_head_dim": null, | |
| "quantization_config": { | |
| "_load_in_4bit": true, | |
| "_load_in_8bit": false, | |
| "bnb_4bit_compute_dtype": "bfloat16", | |
| "bnb_4bit_quant_storage": "uint8", | |
| "bnb_4bit_quant_type": "nf4", | |
| "bnb_4bit_use_double_quant": true, | |
| "llm_int8_enable_fp32_cpu_offload": false, | |
| "llm_int8_has_fp16_weight": false, | |
| "llm_int8_skip_modules": [ | |
| "vae", | |
| "model.vae", | |
| "vae.decoder", | |
| "vae.encoder", | |
| "autoencoder", | |
| "model.autoencoder", | |
| "autoencoder.decoder", | |
| "autoencoder.encoder", | |
| "vision_model", | |
| "model.vision_model", | |
| "vision_aligner", | |
| "model.vision_aligner", | |
| "patch_embed", | |
| "model.patch_embed", | |
| "final_layer", | |
| "model.final_layer", | |
| "time_embed", | |
| "model.time_embed", | |
| "time_embed_2", | |
| "model.time_embed_2", | |
| "timestep_emb", | |
| "model.timestep_emb", | |
| "guidance_emb", | |
| "model.guidance_emb", | |
| "timestep_r_emb", | |
| "model.timestep_r_emb", | |
| "attn.q_proj", | |
| "attn.k_proj", | |
| "attn.v_proj", | |
| "attn.o_proj", | |
| "attn.qkv_proj", | |
| "attn.out_proj", | |
| "self_attn", | |
| "cross_attn" | |
| ], | |
| "llm_int8_threshold": 6.0, | |
| "load_in_4bit": true, | |
| "load_in_8bit": false, | |
| "quant_method": "bitsandbytes" | |
| }, | |
| "rms_norm_eps": 1e-05, | |
| "rope_scaling": { | |
| "alpha": 1.0, | |
| "beta_fast": 32, | |
| "beta_slow": 1, | |
| "factor": 1.0, | |
| "mscale": 1.0, | |
| "mscale_all_dim": 1.0, | |
| "type": "custom" | |
| }, | |
| "rope_theta": 10000.0, | |
| "rope_type": "2d", | |
| "routed_scaling_factor": false, | |
| "skip_cls_token": false, | |
| "text_end_id": 7, | |
| "text_start_id": 6, | |
| "tie_word_embeddings": false, | |
| "topk_group": false, | |
| "transformers_version": "4.57.1", | |
| "use_cache": true, | |
| "use_cla": false, | |
| "use_meanflow": true, | |
| "use_mixed_mlp_moe": true, | |
| "use_mla": false, | |
| "use_qk_norm": true, | |
| "use_rotary_pos_emb": true, | |
| "v_head_dim": null, | |
| "vae": { | |
| "_class_name": "AutoencoderKLConv3D", | |
| "block_out_channels": [ | |
| 128, | |
| 256, | |
| 512, | |
| 1024, | |
| 1024 | |
| ], | |
| "downsample_match_channel": true, | |
| "ffactor_spatial": 16, | |
| "ffactor_temporal": 4, | |
| "in_channels": 3, | |
| "latent_channels": 32, | |
| "layers_per_block": 2, | |
| "out_channels": 3, | |
| "sample_size": 384, | |
| "sample_tsize": 96, | |
| "scaling_factor": 0.562679178327931, | |
| "upsample_match_channel": true | |
| }, | |
| "vae_autocast_dtype": "float16", | |
| "vae_downsample_factor": [ | |
| 16, | |
| 16 | |
| ], | |
| "vae_dtype": "float32", | |
| "vae_type": "hunyuan-image-vae-v1", | |
| "video_end_id": 10, | |
| "video_start_id": 9, | |
| "vit": { | |
| "_attn_implementation": "sdpa", | |
| "attention_dropout": 0.0, | |
| "hidden_act": "gelu_pytorch_tanh", | |
| "hidden_size": 1152, | |
| "intermediate_size": 4304, | |
| "layer_norm_eps": 1e-06, | |
| "num_attention_heads": 16, | |
| "num_channels": 3, | |
| "num_hidden_layers": 27, | |
| "num_patches": 256, | |
| "output_attentions": false, | |
| "output_hidden_states": false, | |
| "patch_size": 16, | |
| "torch_dtype": "float32", | |
| "use_return_dict": true | |
| }, | |
| "vit_add_patchemb_bias": false, | |
| "vit_aligner": { | |
| "depth": 2, | |
| "input_dim": 1152, | |
| "n_embed": 4096, | |
| "projector_type": "mlp_gelu", | |
| "torch_dtype": "float32" | |
| }, | |
| "vit_downsample_factor": [ | |
| 16, | |
| 16 | |
| ], | |
| "vit_input_resolution": 224, | |
| "vit_mapping_type": "resampler", | |
| "vit_norm_type": "fused", | |
| "vit_patch": 1, | |
| "vit_path": null, | |
| "vit_processor": { | |
| "do_convert_rgb": null, | |
| "do_normalize": true, | |
| "do_rescale": true, | |
| "do_resize": true, | |
| "image_mean": [ | |
| 0.5, | |
| 0.5, | |
| 0.5 | |
| ], | |
| "image_processor_type": "Siglip2ImageProcessorFast", | |
| "image_std": [ | |
| 0.5, | |
| 0.5, | |
| 0.5 | |
| ], | |
| "max_num_patches": 1024, | |
| "patch_size": 16, | |
| "processor_class": "Siglip2Processor", | |
| "resample": 2, | |
| "rescale_factor": 0.00392156862745098 | |
| }, | |
| "vit_remove_prenorm": false, | |
| "vit_token": 64, | |
| "vit_type": "siglip2-so400m-patch16-naflex", | |
| "vit_used_rms_norm": false, | |
| "vocab_size": 133120, | |
| "xdrope_section": null | |
| } | |