chatglm3-6b-32k/bert4torch_config.json

{
  "model": "glm2",
  "hidden_act": "swiglu", 
  "hidden_size": 4096,
  "intermediate_size": 13696,
  "layer_norm_eps": 1e-05,
  "max_sequence_length": 32768,
  "num_attention_heads": 32,
  "num_hidden_layers": 28,
  "vocab_size": 65024,
  "segment_vocab_size": 0,
  "multi_query_group_num": 2,
  "skip_init": true,
  "tie_emb_prj_weight": false,
  "eos_token_id": 2,
  "pad_token_id": 2,
  "rmsnorm": true,
  "rope_rank": "adjacent",
  "ntk_alpha": 44.24632364389211,
  "position_encoding_2d": true,
  "flash_attention": true
}