Qwen/Qwen-7B/bert4torch_config.json

{
    "model": "qwen",
    "hidden_act": "silu",
    "pad_token_id": 151643,
    "eos_token_id": 151643,
    "intermediate_size": 11008,
    "initializer_range": 0.02,
    "layer_norm_eps": 1e-05,
    "hidden_size": 4096,
    "num_attention_heads": 32,
    "num_hidden_layers": 32,
    "rope_theta": 10000,
    "scale_attn_weights": true,
    "tie_word_embeddings": false,
    "_attn_implementation": "flash_attn_2",
    "vocab_size": 151936,
    "rope_scaling": {"type": "dynamic_qwen"},
    "use_logn_attn": true,
    "segment_vocab_size": 0,
    "skip_init": true,
    "rope_rank": "updown",
    "max_position": 8192,
    "generation_config": {"tokenizer_config": {"allowed_special": ["<|im_end|>", "<|im_start|>", "<|endoftext|>"], 
                          "skip_special_tokens": true}, "eos_token_id": [151643],
                          "max_length": 8192}
}