config.json

{
  "architectures": [
    "VLMambaCLIPLMHeadModel"
  ],
  "d_model": 2048,
  "fused_add_norm": true,
  "hidden_size": 2048,
  "image_size": 336,
  "n_layer": 48,
  "pad_vocab_size_multiple": 8,
  "residual_in_fp32": true,
  "rms_norm": true,
  "select_feature": "patch",
  "select_layer": -2,
  "ssm_cfg": {},
  "torch_dtype": "bfloat16",
  "transformers_version": "4.37.1",
  "vision_encoder_name": "timm/eva02_large_patch14_clip_336",
  "vocab_size": 50280
}