Safetensors
mixtral
Kaz-K's picture
first commit
d15dc11 verified
raw
history blame contribute delete
930 Bytes
{
"_name_or_path": "/home/sip3-aizawa-test/shared/kkobayashi/002_SIP-pre2-8x13b_CPT_MedCv2OP/checkpoints_hf/iter_0010551",
"architectures": [
"MixtralForCausalLM"
],
"attention_bias": false,
"attention_dropout": 0.0,
"bos_token_id": 1,
"eos_token_id": 2,
"head_dim": 128,
"hidden_act": "silu",
"hidden_size": 5120,
"initializer_range": 0.02,
"intermediate_size": 13824,
"max_position_embeddings": 4096,
"mlp_bias": false,
"model_type": "mixtral",
"num_attention_heads": 40,
"num_experts_per_tok": 2,
"num_hidden_layers": 40,
"num_key_value_heads": 40,
"num_local_experts": 8,
"output_router_logits": false,
"rms_norm_eps": 1e-05,
"rope_theta": 10000,
"router_aux_loss_coef": 0.01,
"router_jitter_noise": 0.0,
"sliding_window": null,
"tie_word_embeddings": false,
"torch_dtype": "float32",
"transformers_version": "4.47.0",
"use_cache": true,
"vocab_size": 99584
}