| { |
| "architectures": [ |
| "LLaVAOneVision1_5_ForConditionalGeneration" |
| ], |
| "auto_map": { |
| "AutoConfig": "lmms-lab/LLaVA-OneVision-1.5-8B-Instruct--configuration_llavaonevision1_5.Llavaonevision1_5Config", |
| "AutoModel": "lmms-lab/LLaVA-OneVision-1.5-8B-Instruct--modeling_llavaonevision1_5.LLaVAOneVision1_5_ForConditionalGeneration", |
| "AutoModelForCausalLM": "lmms-lab/LLaVA-OneVision-1.5-8B-Instruct--modeling_llavaonevision1_5.LLaVAOneVision1_5_ForConditionalGeneration" |
| }, |
| "dtype": "bfloat16", |
| "image_token_id": 151655, |
| "model_type": "llavaonevision1_5", |
| "text_config": { |
| "attention_bias": false, |
| "attention_dropout": 0.0, |
| "head_dim": 32, |
| "hidden_act": "silu", |
| "hidden_size": 8, |
| "image_token_id": null, |
| "initializer_range": 0.02, |
| "intermediate_size": 64, |
| "layer_types": [ |
| "full_attention", |
| "full_attention" |
| ], |
| "max_position_embeddings": 32768, |
| "max_window_layers": 2, |
| "model_type": "LLaVAOneVision1_5_text", |
| "num_attention_heads": 8, |
| "num_hidden_layers": 2, |
| "num_key_value_heads": 4, |
| "rms_norm_eps": 1e-06, |
| "rope_scaling": null, |
| "rope_theta": 1000000.0, |
| "sliding_window": null, |
| "use_cache": true, |
| "use_sliding_window": false, |
| "video_token_id": null, |
| "vocab_size": 151936 |
| }, |
| "transformers_version": "4.57.0.dev0", |
| "video_token_id": 151656, |
| "vision_config": { |
| "depth": 2, |
| "embed_dim": 128, |
| "hidden_act": "gelu", |
| "hidden_size": 128, |
| "in_channels": 3, |
| "initializer_range": 0.02, |
| "intermediate_size": 256, |
| "layer_norm_eps": 1e-05, |
| "model_type": "rice_vit", |
| "num_heads": 4, |
| "num_hidden_layers": 2, |
| "patch_size": 14, |
| "spatial_merge_size": 2, |
| "temporal_patch_size": 1, |
| "text_hidden_size": 8 |
| }, |
| "vocab_size": 151936 |
| } |
