zm_config.json

{
  "library_name": "zeromodels",
  "zeromodels_version": "1.2.2",
  "model_module": "zeromodels.models.qwen2_5_vl",
  "model_class": "Qwen2_5VLConditionalGenerate",
  "variant": "qwen2.5-vl-7b-instruct",
  "weights": "model.weights.json",
  "schema_version": 2,
  "weight_dtype": "bfloat16",
  "model_type": "qwen2_5_vl",
  "text_config": {
    "vocab_size": 152064,
    "embed_dim": 3584,
    "mlp_dim": 18944,
    "num_layers": 28,
    "num_heads": 28,
    "num_kv_heads": 4,
    "norm_eps": 1e-06,
    "rope_theta": 1000000.0,
    "mrope_section": [
      16,
      24,
      24
    ],
    "tie_embeddings": false
  },
  "vision_config": {
    "depth": 32,
    "embed_dim": 1280,
    "mlp_dim": 3420,
    "num_heads": 16,
    "out_dim": 3584,
    "window_size": 112,
    "fullatt_block_indexes": [
      7,
      15,
      23,
      31
    ],
    "tokens_per_second": 2,
    "patch_size": 14,
    "spatial_merge_size": 2,
    "temporal_patch_size": 2,
    "in_channels": 3
  },
  "image_token_id": 151655,
  "video_token_id": 151656,
  "vision_start_token_id": 151652,
  "vision_end_token_id": 151653
}