Files
minimind3-ascend-moe/config.json

36 lines
869 B
JSON
Raw Permalink Normal View History

{
"architectures": [
"Qwen3MoeForCausalLM"
],
"attention_bias": false,
"attention_dropout": 0.0,
"decoder_sparse_step": 1,
"dtype": "float16",
"head_dim": 96,
"hidden_act": "silu",
"hidden_size": 768,
"initializer_range": 0.02,
"intermediate_size": 2432,
"max_position_embeddings": 8192,
"mlp_only_layers": [],
"model_type": "qwen3_moe",
"moe_intermediate_size": 2432,
"norm_topk_prob": true,
"num_attention_heads": 8,
"num_experts": 4,
"num_experts_per_tok": 1,
"num_hidden_layers": 8,
"num_key_value_heads": 4,
"output_router_logits": false,
"rms_norm_eps": 1e-06,
"rope_scaling": null,
"rope_theta": 1000000.0,
"router_aux_loss_coef": 0.001,
"sliding_window": null,
"tie_word_embeddings": true,
"transformers_version": "4.57.6",
"use_cache": true,
"use_sliding_window": false,
"vocab_size": 6400
}