diff --git a/inference/configs/config_kimi_k2.json b/inference/configs/config_kimi_k2.json new file mode 100644 index 0000000..2b48ce4 --- /dev/null +++ b/inference/configs/config_kimi_k2.json @@ -0,0 +1,26 @@ +{ + "vocab_size": 163840, + "dim": 7168, + "inter_dim": 18432, + "moe_inter_dim": 2048, + "n_layers": 61, + "n_dense_layers": 1, + "n_heads": 64, + "n_routed_experts": 384, + "n_shared_experts": 1, + "n_activated_experts": 8, + "n_expert_groups": 1, + "n_limited_groups": 1, + "route_scale": 2.827, + "score_func": "sigmoid", + "q_lora_rank": 1536, + "kv_lora_rank": 512, + "qk_nope_head_dim": 128, + "qk_rope_head_dim": 64, + "v_head_dim": 128, + "dtype": "fp8", + "rope_theta": 50000.0, + "rope_factor": 64, + "beta_fast": 1, + "beta_slow": 1 +}