text_encoder/config.json

{
  "_class_name": "MingImageTextEncoder",
  "_diffusers_version": "0.41.0.dev0",
  "auto_map": {
    "AutoModel": "text_encoder_ming_image.MingImageTextEncoder"
  },
  "first_k_dense_replace": 1,
  "head_dim": 128,
  "hidden_size": 2048,
  "image_end_token": 157159,
  "image_patch_token": 157157,
  "image_start_token": 157158,
  "intermediate_size": 5120,
  "moe_intermediate_size": 512,
  "mrope_section": [
    8,
    12,
    12
  ],
  "n_group": 8,
  "num_attention_heads": 16,
  "num_experts": 256,
  "num_experts_per_tok": 8,
  "num_hidden_layers": 20,
  "num_key_value_heads": 4,
  "num_query_tokens": 256,
  "num_shared_experts": 1,
  "pad_token_id": 156892,
  "partial_rotary_factor": 0.5,
  "rms_norm_eps": 1e-06,
  "rope_theta": 600000,
  "rope_time_scale": 2.0,
  "routed_scaling_factor": 2.5,
  "spatial_merge_size": 2,
  "topk_group": 4,
  "vocab_size": 157184
}