32 lines
821 B
JSON
32 lines
821 B
JSON
{
|
|
"architectures": [
|
|
"KohakuForCausalLM"
|
|
],
|
|
"model_type": "kohaku",
|
|
"auto_map": {
|
|
"AutoConfig": "configuration_kohaku.KohakuConfig",
|
|
"AutoModel": "modeling_kohaku.KohakuModel",
|
|
"AutoModelForCausalLM": "modeling_kohaku.KohakuForCausalLM"
|
|
},
|
|
"vocab_size": 65536,
|
|
"hidden_size": 768,
|
|
"num_hidden_layers": 16,
|
|
"num_attention_heads": 12,
|
|
"num_key_value_heads": 2,
|
|
"head_dim": 64,
|
|
"intermediate_size": 2048,
|
|
"moe_intermediate_size": 384,
|
|
"n_routed_experts": 64,
|
|
"n_shared_experts": 1,
|
|
"num_experts_per_tok": 8,
|
|
"first_k_dense": 1,
|
|
"norm_topk_prob": true,
|
|
"routed_scaling_factor": 1.0,
|
|
"scoring_func": "sigmoid",
|
|
"max_position_embeddings": 4096,
|
|
"rope_theta": 100000.0,
|
|
"rms_norm_eps": 1e-06,
|
|
"qk_norm": true,
|
|
"tie_word_embeddings": false,
|
|
"dtype": "float16"
|
|
} |