| { | |
| "dropout": 0.0, | |
| "bos_token_id": 151644, | |
| "eos_token_id": 151645, | |
| "pad_token_id": 151643, | |
| "hidden_act": "silu", | |
| "hidden_size": 768, | |
| "num_hidden_layers": 8, | |
| "max_position_embeddings": 4096, | |
| "vocab_size": 6400, | |
| "rms_norm_eps": 1e-06, | |
| "rope_theta": 50000.0, | |
| "rope_scaling": null, | |
| "dtype": "bfloat16", | |
| "self_distill": true, | |
| "intermediate_size": 1536, | |
| "expert_intermediate_size": 768, | |
| "n_routed_experts": 0, | |
| "moe_topk": 2, | |
| "score_func": "softmax", | |
| "n_shared_experts": 0, | |
| "top_k_layer_dense": 8, | |
| "aux_loss_alpha": 0.02, | |
| "seq_aux": false, | |
| "norm_topk_prob": true, | |
| "noisy_expert": 0.0, | |
| "moe_backend": "compact", | |
| "router_bias_enabled": true, | |
| "router_bias_update_rate": 0.001, | |
| "router_bias_clamp": 5.0, | |
| "num_heads": 6, | |
| "mla_kv_lora_rank": 128, | |
| "mla_qk_nope_head_dim": 64, | |
| "mla_qk_rope_head_dim": 64, | |
| "mla_attn_impl": "absorb", | |
| "qkv_lora": false, | |
| "model_type": "ynet3", | |
| "architectures": [ | |
| "YForCausalLM3" | |
| ], | |
| "auto_map": { | |
| "AutoConfig": "configuration_ymodel3.YConfig3", | |
| "AutoModelForCausalLM": "modeling_ymodel3.YForCausalLM3" | |
| } | |
| } |