Qwen-14B-Chat/bert4torch_config.json

{
    "model": "qwen",
    "hidden_act": "silu",
    "intermediate_size": 13696,
    "initializer_range": 0.02,
    "layer_norm_eps": 1e-06,
    "hidden_size": 5120,
    "num_attention_heads": 40,
    "num_hidden_layers": 40,
    "rotary_emb_base": 10000,
    "rotary_pct": 1.0,
    "scale_attn_weights": true,
    "seq_length": 2048,
    "tie_word_embeddings": false,
    "use_cache": true,
    "flash_attention": "flash_attn_2",
    "vocab_size": 152064,
    "use_dynamic_ntk": true,
    "use_logn_attn": true,
    "segment_vocab_size": 0,
    "skip_init": true,
    "rope_rank": "updown",
    "max_position_embeddings": 8192
}