Qwen/Qwen2-0.5B-Instruct/bert4torch_config.json

{
    "model": "qwen2",
    "hidden_act": "silu",
	"attention_probs_dropout_prob": 0.0,
    "bos_token_id": 151643,
    "eos_token_id": 151645,
    "intermediate_size": 4864,
    "initializer_range": 0.02,
    "layer_norm_eps": 1e-06,
    "hidden_size": 896,
    "num_attention_heads": 14,
    "num_hidden_layers": 24,
	"num_key_value_heads": 2,
    "rope_theta": 1000000.0,
    "tie_word_embeddings": true,
	"torch_dtype": "bfloat16",
    "_attn_implementation": "sdpa",
	"is_causal": true,
    "vocab_size": 151936,
    "segment_vocab_size": 0,
    "skip_init": true,
    "rope_rank": "updown",
	"max_position_embeddings": 32768,
	"sliding_window": 32768,
	"max_window_layers": 24,
    "convert_lm_logits_dtype": "float32",
    "generation_config": {"tokenizer_config": {"skip_special_tokens": true}, "eos_token_id": [151643, 151645],
                          "max_length": 32768}
}