config.json

{
  "architectures": [
    "GlmOcrForConditionalGeneration"
  ],
  "dtype": "bfloat16",
  "image_end_token_id": 59257,
  "image_start_token_id": 59256,
  "image_token_id": 59280,
  "model_type": "glm_ocr",
  "text_config": {
    "attention_bias": false,
    "attention_dropout": 0.0,
    "dtype": "bfloat16",
    "eos_token_id": [
      59246,
      59253
    ],
    "head_dim": 32,
    "hidden_act": "silu",
    "hidden_size": 8,
    "initializer_range": 0.02,
    "intermediate_size": 64,
    "max_position_embeddings": 131072,
    "model_type": "glm_ocr_text",
    "num_attention_heads": 8,
    "num_hidden_layers": 2,
    "num_key_value_heads": 4,
    "num_nextn_predict_layers": 1,
    "pad_token_id": 59246,
    "rms_norm_eps": 1e-05,
    "rope_parameters": {
      "mrope_section": [
        4,
        4,
        8
      ],
      "partial_rotary_factor": 1.0,
      "rope_theta": 10000,
      "rope_type": "default"
    },
    "tie_word_embeddings": false,
    "use_cache": true,
    "vocab_size": 59392
  },
  "tie_word_embeddings": false,
  "transformers_version": "5.2.0.dev0",
  "video_end_token_id": 59259,
  "video_start_token_id": 59258,
  "video_token_id": 59281,
  "vision_config": {
    "attention_bias": true,
    "attention_dropout": 0.0,
    "depth": 2,
    "hidden_act": "silu",
    "hidden_dropout_prob": 0.0,
    "hidden_size": 32,
    "image_size": 336,
    "in_channels": 3,
    "initializer_range": 0.02,
    "intermediate_size": 64,
    "model_type": "glm_ocr_vision",
    "num_heads": 1,
    "out_hidden_size": 8,
    "patch_size": 14,
    "rms_norm_eps": 1e-05,
    "spatial_merge_size": 2,
    "temporal_patch_size": 2
  }
}