EgoQA-GeLM-7B/config.json

{
  "_name_or_path": "./checkpoints/vicuna-v1-3-7b",
  "architectures": [
    "GelmLlamaForCausalLM"
  ],
  "attention_bias": false,
  "bos_token_id": 1,
  "eos_token_id": 2,
  "freeze_mm_mlp_adapter": false,
  "grounding_args": {
    "d_gnd": 1024,
    "d_proj": 1024,
    "gnd_enc_layers": 3
  },
  "hidden_act": "silu",
  "hidden_size": 4096,
  "image_aspect_ratio": "square",
  "image_grid_pinpoints": null,
  "initializer_range": 0.02,
  "input_type": "feature",
  "intermediate_size": 11008,
  "max_position_embeddings": 2048,
  "mm_hidden_size": 768,
  "mm_use_im_patch_token": false,
  "mm_use_im_start_end": false,
  "mm_vision_select_feature": "patch",
  "mm_vision_select_layer": -2,
  "mm_vision_tower": "/remote-home/share/huggingface/clip-vit-large-patch14/",
  "model_type": "gelm",
  "num_attention_heads": 32,
  "num_frames": 180,
  "num_hidden_layers": 32,
  "num_key_value_heads": 32,
  "pad_token_id": 0,
  "pretraining_tp": 1,
  "rms_norm_eps": 1e-06,
  "rope_scaling": null,
  "rope_theta": 10000.0,
  "saliency_token_ids": [
    32010,
    32011
  ],
  "special_token_ids": [
    32000,
    32001,
    32002,
    32003,
    32004,
    32005,
    32006,
    32007,
    32008,
    32009
  ],
  "tie_word_embeddings": false,
  "torch_dtype": "bfloat16",
  "transformers_version": "4.34.0",
  "tune_mm_mlp_adapter": false,
  "use_cache": true,
  "use_mm_proj": true,
  "vocab_size": 32012
}