config.json

{
  "architectures": [
    "GlmImageTextModel"
  ],
  "attention_bias": true,
  "attention_dropout": 0.0,
  "bos_token_id": 16384,
  "dtype": "float32",
  "eos_token_id": 16385,
  "hidden_act": "silu",
  "hidden_size": 16,
  "initializer_range": 0.02,
  "intermediate_size": 22,
  "max_position_embeddings": 512,
  "model_type": "glm_image_text",
  "n_group": 1,
  "n_routed_experts": 8,
  "n_shared_experts": 1,
  "num_attention_heads": 2,
  "num_experts_per_tok": 8,
  "num_hidden_layers": 2,
  "num_key_value_heads": 1,
  "output_channels": 64,
  "pad_token_id": 16385,
  "rms_norm_eps": 1e-05,
  "rope_parameters": {
    "mrope_section": [
      2,
      1,
      1
    ],
    "rope_theta": 10000,
    "rope_type": "default",
    "type": "default"
  },
  "tie_word_embeddings": true,
  "topk_group": 1,
  "transformers_version": "5.16.0.dev0",
  "use_cache": true,
  "vision_vocab_size": 99,
  "vocab_size": 167864
}