{ | |
"architectures": [ | |
"DeepSeekForCausalLM" | |
], | |
"model_type": "deepseek", | |
"vocab_size": 50256, | |
"hidden_size": 1024, | |
"num_hidden_layers": 6, | |
"num_dense_layers": 1, | |
"num_attention_heads": 8, | |
"intermediate_size": 4096, | |
"moe_intermediate_size": 704, | |
"num_routed_experts": 4, | |
"num_shared_experts": 2, | |
"num_activated_experts": 2, | |
"max_position_embeddings": 256, | |
"q_lora_rank": 0, | |
"kv_lora_rank": 256, | |
"qk_nope_head_dim": 64, | |
"qk_rope_head_dim": 32, | |
"v_head_dim": 64, | |
"original_seq_len": 512, | |
"rope_theta": 10000.0, | |
"rope_factor": 40, | |
"beta_fast": 32, | |
"beta_slow": 1, | |
"mscale": 1.0, | |
"rms_norm_eps": 0.001, | |
"initializer_range": 0.02, | |
"use_cache": true, | |
"pad_token_id": 0, | |
"bos_token_id": 2, | |
"eos_token_id": 3, | |
"tie_word_embeddings": false, | |
"torch_dtype": "float32", | |
"auto_map": { | |
"AutoConfig": "modeling_deepseek.DeepSeekConfig", | |
"AutoModel": "modeling_deepseek.DeepSeekModel", | |
"AutoModelForCausalLM": "modeling_deepseek.DeepSeekForCausalLM" | |
} | |
} |