opensourcerelease/DeepSeek-R1-bf16
1754
1{2 "architectures": [3 "DeepseekV3ForCausalLM"4 ],5 "attention_bias": false,6 "attention_dropout": 0.0,7 "auto_map": {8 "AutoConfig": "configuration_deepseek.DeepseekV3Config",9 "AutoModel": "modeling_deepseek.DeepseekV3Model",10 "AutoModelForCausalLM": "modeling_deepseek.DeepseekV3ForCausalLM"11 },12 "aux_loss_alpha": 0.001,13 "bos_token_id": 0,14 "eos_token_id": 1,15 "ep_size": 1,16 "first_k_dense_replace": 3,17 "hidden_act": "silu",18 "hidden_size": 7168,19 "initializer_range": 0.02,20 "intermediate_size": 18432,21 "kv_lora_rank": 512,22 "max_position_embeddings": 163840,23 "model_type": "deepseek_v3",24 "moe_intermediate_size": 2048,25 "moe_layer_freq": 1,26 "n_group": 8,27 "n_routed_experts": 256,28 "n_shared_experts": 1,29 "norm_topk_prob": true,30 "num_attention_heads": 128,31 "num_experts_per_tok": 8,32 "num_hidden_layers": 61,33 "num_key_value_heads": 128,34 "num_nextn_predict_layers": 1,35 "pretraining_tp": 1,36 "q_lora_rank": 1536,37 "qk_nope_head_dim": 128,38 "qk_rope_head_dim": 64,39 "rms_norm_eps": 1e-06,40 "rope_scaling": {41 "beta_fast": 32,42 "beta_slow": 1,43 "factor": 40,44 "mscale": 1.0,45 "mscale_all_dim": 1.0,46 "original_max_position_embeddings": 4096,47 "type": "yarn"48 },49 "rope_theta": 10000,50 "routed_scaling_factor": 2.5,51 "scoring_func": "sigmoid",52 "seq_aux": true,53 "tie_word_embeddings": false,54 "topk_group": 4,55 "topk_method": "noaux_tc",56 "torch_dtype": "bfloat16",57 "transformers_version": "4.46.3",58 "use_cache": true,59 "v_head_dim": 128,60 "vocab_size": 12928061}62 