{
  "anyres": false,
  "architectures": [
    "HyperCLOVAXVisionV2ForConditionalGeneration"
  ],
  "eos_token_id": 128001,
  "hidden_size": 5120,
  "ignore_index": -100,
  "image_end_token_id": 128057,
  "image_start_token_id": 128056,
  "image_token_id": 128060,
  "max_num_grids": -1,
  "audio_projector_type": "mlp",
  "vision_projector_type": "linear",
  "model_type": "hyperclovax_vision_v2",
  "num_queries_vis_abstractor": -1,
  "pad_token_id": 0,
  "possible_resolutions": [],
  "proj_pos_emb": true,
  "proj_prenorm": false,
  "text_config": {
    "architectures": [
      "HyperCLOVAXForCausalLM"
    ],
    "attention_bias": false,
    "attention_dropout": 0.0,
    "attention_multiplier": 0.08838834764831845,
    "bos_token_id": 128000,
    "embedding_multiplier": 1.0,
    "eos_token_id": 128001,
    "head_dim": 128,
    "hidden_act": "silu",
    "hidden_size": 5120,
    "initializer_range": 0.006,
    "intermediate_size": 24192,
    "logits_scaling": 1.0,
    "max_position_embeddings": 131072,
    "mlp_bias": false,
    "model_type": "hyperclovax",
    "num_attention_heads": 40,
    "num_hidden_layers": 72,
    "num_key_value_heads": 8,
    "pad_token_id": 0,
    "pretraining_tp": 1,
    "resid_pdrop": 0.2,
    "residual_multiplier": 1.0,
    "rms_norm_eps": 1e-05,
    "rope_scaling": null,
    "rope_theta": 50000000,
    "tie_word_embeddings": false,
    "use_cache": true,
    "use_post_norm": false,
    "vocab_size": 128256
  },
  "torch_dtype": "bfloat16",
  "transformers_version": "4.58.0",
  "unpad": false,
  "use_1x1_grid": false,
  "video_first_last_frames_slows": null,
  "video_max_num_frames": null,
  "video_num_queries_fast": null,
  "video_num_queries_slow": null,
  "video_end_token_id": 128059,
  "video_start_token_id": 128058,
  "video_token_id": 128061,
  "vision_config": {
    "anyres": false,
    "architectures": [
      "Qwen2_5_VisionTransformerPretrainedModel"
    ],
    "depth": 32,
    "fullatt_block_indexes": [
      7,
      15,
      23,
      31
    ],
    "hidden_act": "silu",
    "hidden_size": 1280,
    "in_channels": 3,
    "in_chans": 3,
    "initializer_range": 0.02,
    "intermediate_size": 3456,
    "max_num_grids": -1,
    "model_type": "qwen2_5_vl_vision",
    "num_heads": 16,
    "out_hidden_size": 5120,
    "patch_size": 14,
    "spatial_merge_size": 2,
    "spatial_patch_size": 14,
    "temporal_patch_size": 2,
    "tokens_per_second": 2,
    "window_size": 112
  },
  "vision_feature_layer": -2,
  "vision_input_chunk_size": null
}