| { | |
| "architectures": [ | |
| "DeepseekForCausalLM" | |
| ], | |
| "attention_bias": false, | |
| "attention_dropout": 0.0, | |
| "auto_map": { | |
| "AutoConfig": "configuration_deepseek.DeepseekConfig", | |
| "AutoModel": "modeling_deepseek.DeepseekModel", | |
| "AutoModelForCausalLM": "modeling_deepseek.DeepseekForCausalLM" | |
| }, | |
| "bos_token_id": 100000, | |
| "eos_token_id": 100001, | |
| "first_k_dense_replace": 1, | |
| "hidden_act": "silu", | |
| "hidden_size": 2048, | |
| "initializer_range": 0.02, | |
| "intermediate_size": 10944, | |
| "max_position_embeddings": 4096, | |
| "model_type": "deepseek", | |
| "moe_intermediate_size": 1408, | |
| "moe_layer_freq": 1, | |
| "n_routed_experts": 64, | |
| "n_shared_experts": 2, | |
| "norm_topk_prob": false, | |
| "num_attention_heads": 16, | |
| "num_experts_per_tok": 6, | |
| "num_hidden_layers": 28, | |
| "num_key_value_heads": 16, | |
| "pretraining_tp": 1, | |
| "rms_norm_eps": 1e-06, | |
| "rope_scaling": null, | |
| "rope_theta": 10000, | |
| "scoring_func": "softmax", | |
| "tie_word_embeddings": false, | |
| "torch_dtype": "bfloat16", | |
| "transformers_version": "4.36.0", | |
| "use_cache": true, | |
| "vocab_size": 102400 | |
| } | |