| { |
| "_name_or_path": "meta-llama/Llama-3.2-3B-Instruct", |
| "add_faster_video": false, |
| "add_time_instruction": false, |
| "architectures": [ |
| "LlavaLlamaForCausalLM" |
| ], |
| "attention_bias": false, |
| "attention_dropout": 0.0, |
| "bos_token_id": 128000, |
| "eos_token_id": [ |
| 128001, |
| 128008, |
| 128009 |
| ], |
| "faster_token_stride": 10, |
| "force_sample": false, |
| "frame_dropout": null, |
| "head_dim": 128, |
| "hidden_act": "silu", |
| "hidden_size": 3072, |
| "image_aspect_ratio": "square", |
| "image_crop_resolution": null, |
| "image_grid_pinpoints": null, |
| "image_split_resolution": null, |
| "initializer_range": 0.02, |
| "intermediate_size": 8192, |
| "label_top_k": 5, |
| "max_position_embeddings": 131072, |
| "max_text_pe": 2048, |
| "mlp_bias": false, |
| "mm_hidden_size": 1152, |
| "mm_newline_position": "grid", |
| "mm_patch_merge_type": "flat", |
| "mm_perceiver_depth": 2, |
| "mm_perceiver_ff_mult": 4, |
| "mm_perceiver_heads": 2, |
| "mm_perceiver_latents": 256, |
| "mm_perceiver_pretrained": null, |
| "mm_projector_lr": null, |
| "mm_projector_type": "mlp2x_gelu", |
| "mm_resampler_type": "perceiverv2", |
| "mm_selector_input": "cls", |
| "mm_selector_size": 2304, |
| "mm_selector_type": "selector_attn", |
| "mm_spatial_pool_mode": "bilinear", |
| "mm_spatial_pool_stride": null, |
| "mm_text_size": 1152, |
| "mm_text_tower": "google/siglip-so400m-patch14-384", |
| "mm_text_tower_lr": 2e-06, |
| "mm_tunable_parts": "mm_mlp_adapter,mm_vision_resampler,mm_mlp_selector,mm_text_tower,mm_vision_tower,mm_language_model", |
| "mm_use_im_patch_token": false, |
| "mm_use_im_start_end": false, |
| "mm_vision_select_feature": "cls_patch", |
| "mm_vision_select_layer": -2, |
| "mm_vision_tower": "google/siglip-so400m-patch14-384", |
| "mm_vision_tower_lr": 2e-06, |
| "model_type": "llama", |
| "num_attention_heads": 24, |
| "num_hidden_layers": 28, |
| "num_key_value_heads": 8, |
| "patch_dropout_rate": 0.4, |
| "pos_skipping_range": 4096, |
| "pretraining_tp": 1, |
| "rms_norm_eps": 1e-05, |
| "rope_scaling": { |
| "factor": 32.0, |
| "high_freq_factor": 4.0, |
| "low_freq_factor": 1.0, |
| "original_max_position_embeddings": 8192, |
| "rope_type": "llama3" |
| }, |
| "rope_theta": 500000.0, |
| "selector_activation": "prelu", |
| "selector_d_model": 1152, |
| "selector_dim_feedforward": 1152, |
| "selector_dropout": 0.0, |
| "selector_is_double": false, |
| "selector_nhead": 1, |
| "selector_normalize_before": false, |
| "selector_num_decoder_layers": 1, |
| "selector_num_encoder_layers": 2, |
| "selector_num_queries": 5, |
| "text_tower_pretrained": null, |
| "tie_word_embeddings": true, |
| "tokenizer_model_max_length": 8192, |
| "tokenizer_padding_side": "right", |
| "torch_dtype": "bfloat16", |
| "train_step": "sft", |
| "transformers_version": "4.45.2", |
| "use_cache": false, |
| "use_mm_proj": true, |
| "use_pos_skipping": false, |
| "vision_tower_pretrained": null, |
| "vocab_size": 128256, |
| "weight_ce": 1.0 |
| } |
|
|
