glm-4-9b/bert4torch_config.json

{
	"model": "glm2",
    "template": "glm4",
    "hidden_act": "swiglu",
    "hidden_size": 4096,
    "intermediate_size": 13696,
    "layer_norm_eps": 1.5625e-07,
    "max_sequence_length": 131072,
    "num_attention_heads": 32,
    "num_hidden_layers": 40,
    "vocab_size": 151552,
    "segment_vocab_size": 0,
    "num_key_value_heads": 2,
    "skip_init": true,
    "tie_emb_prj_weight": false,
    "rmsnorm": true,
    "rope_rank": "adjacent",
    "position_encoding_2d": true,
    "flash_attention": true,
    "eos_token_id": [151329, 151336, 151338],
    "pad_token_id": 151329,
    "generation_config": {"tokenizer_config": {"skip_special_tokens": true}, 
                          "eos_token_id": [151329, 151336, 151338], "max_length": 131072}
}