| { |
| "architectures": [ |
| "Param2MoEForCausalLM" |
| ], |
| "attention_dropout": 0.0, |
| "auto_map": { |
| "AutoConfig": "configuration_param2moe.Param2MoEConfig", |
| "AutoModel": "modeling_param2moe.Param2MoEModel", |
| "AutoModelForCausalLM": "modeling_param2moe.Param2MoEForCausalLM" |
| }, |
| "model_type": "param2moe", |
| "num_hidden_layers": 21, |
| "hidden_size": 2048, |
| "intermediate_size": 9216, |
| "eos_token_id": 3, |
| "pad_token_id": 0, |
| "first_k_dense_replace": 1, |
| "hidden_act": "silu", |
| "max_position_embeddings": 4096, |
| "moe_intermediate_size": 2048, |
| "norm_topk_prob": true, |
| "num_experts_per_tok": 6, |
| "num_attention_heads": 32, |
| "n_routed_experts": 64, |
| "num_key_value_heads": 8, |
| "rope_theta": 10000.0, |
| "tie_word_embeddings": true, |
| "rms_norm_eps": 1e-06, |
| "head_dim": 64, |
| "n_shared_experts": 2, |
| "use_cache": true, |
| "attention_bias": false, |
| "embedding_dropout": 0.0, |
| "output_dropout": 0.0, |
| "vocab_size": 128000, |
| "partial_rotary_factor": 1.0, |
| "router_dtype": "fp32", |
| "routed_scaling_factor": 2.5, |
| "n_group": 1, |
| "topk_group": 1, |
| "use_qk_norm": true, |
| "scoring_func": "sigmoid" |
| } |