config.json

{
  "architectures": [
    "VLMForCausalLM"
  ],
  "assistant_token": null,
  "attention_bias": false,
  "attention_dropout": 0.0,
  "auto_map": {
    "AutoConfig": "configuration_vlm.VLMConfig",
    "AutoModel": "modeling_vlm.VLMForCausalLM"
  },
  "bos_token_id": 1,
  "connector_config": {
    "model_type": "connector",
    "name": "mlp_2_gelu",
    "type": "mlp"
  },
  "eos_token_id": 2,
  "head_dim": 128,
  "hf_name": "lmsys/vicuna-7b-v1.5",
  "hidden_act": "silu",
  "hidden_size": 4096,
  "ignore_index": -100,
  "image_end_token": "<im_end>",
  "image_patch_token": "<im_patch>",
  "image_start_token": "<im_start>",
  "image_token": "<image>",
  "image_token_index": -200,
  "initializer_range": 0.02,
  "intermediate_size": 11008,
  "lazy_load": false,
  "max_position_embeddings": 4096,
  "max_seq_length": 2048,
  "mlp_bias": false,
  "model_type": "vlm",
  "num_attention_heads": 32,
  "num_hidden_layers": 32,
  "num_key_value_heads": 32,
  "pad_token_id": 0,
  "padding_side": "left",
  "pretraining_tp": 1,
  "rms_norm_eps": 1e-05,
  "rope_scaling": null,
  "rope_theta": 10000.0,
  "system_token": null,
  "tie_word_embeddings": false,
  "torch_dtype": "bfloat16",
  "transformers_version": "4.51.3",
  "use_cache": false,
  "use_image_patch_token": false,
  "use_start_end_tokens": false,
  "user_token": null,
  "vision_config": {
    "attention_dropout": 0.0,
    "dropout": 0.0,
    "hf_name": "openai/clip-vit-large-patch14-336",
    "hidden_act": "quick_gelu",
    "hidden_size": 1024,
    "image_size": 336,
    "initializer_factor": 1.0,
    "initializer_range": 0.02,
    "intermediate_size": 4096,
    "layer_norm_eps": 1e-05,
    "model_type": "vision_model",
    "num_attention_heads": 16,
    "num_channels": 3,
    "num_hidden_layers": 24,
    "output_layer": -2,
    "patch_size": 14,
    "projection_dim": 768,
    "use_cls_token": false
  },
  "vocab_size": 32000
}