| { |
| "architectures": [ |
| "KimiLinearForCausalLM" |
| ], |
| "attention_bias": false, |
| "attention_dropout": 0.0, |
| "bos_token_id": 163584, |
| "dtype": "bfloat16", |
| "eos_token_id": 163586, |
| "head_dim": 8, |
| "hidden_act": "silu", |
| "hidden_size": 64, |
| "initializer_range": 0.02, |
| "intermediate_size": 128, |
| "kv_lora_rank": 16, |
| "layer_types": [ |
| "linear_attention", |
| "linear_attention", |
| "linear_attention", |
| "full_attention" |
| ], |
| "linear_conv_kernel_dim": 4, |
| "linear_head_dim": 16, |
| "linear_num_heads": 4, |
| "max_position_embeddings": 1024, |
| "mlp_layer_types": [ |
| "dense", |
| "sparse", |
| "sparse", |
| "sparse" |
| ], |
| "model_type": "kimi_linear", |
| "moe_intermediate_size": 32, |
| "n_group": 1, |
| "num_experts": 8, |
| "n_shared_experts": 1, |
| "norm_topk_prob": true, |
| "num_attention_heads": 4, |
| "num_experts_per_tok": 2, |
| "num_hidden_layers": 4, |
| "num_key_value_heads": 4, |
| "pad_token_id": 163839, |
| "pretraining_tp": 1, |
| "q_lora_rank": null, |
| "qk_head_dim": 24, |
| "qk_nope_head_dim": 16, |
| "qk_rope_head_dim": 8, |
| "rms_norm_eps": 1e-05, |
| "routed_scaling_factor": 2.446, |
| "tie_word_embeddings": false, |
| "topk_group": 1, |
| "transformers_version": "5.16.0.dev0", |
| "use_cache": true, |
| "v_head_dim": 16, |
| "vocab_size": 163840 |
| } |
|
|
