Files
yejin-korean-3b-v2-base/config.json
ModelHub XC a66872ab53 初始化项目,由ModelHub XC社区提供模型
Model: cooler8/yejin-korean-3b-v2-base
Source: Original Platform
2026-09-28 06:49:18 +08:00

33 lines
717 B
JSON

{
"architectures": [
"Qwen3ForCausalLM"
],
"model_type": "qwen3",
"attention_bias": false,
"attention_dropout": 0.0,
"bos_token_id": 1,
"eos_token_id": [
7,
2
],
"pad_token_id": 3,
"hidden_act": "silu",
"hidden_size": 3072,
"intermediate_size": 8192,
"num_hidden_layers": 28,
"num_attention_heads": 24,
"num_key_value_heads": 8,
"head_dim": 128,
"max_position_embeddings": 4096,
"initializer_range": 0.02,
"rms_norm_eps": 1e-05,
"rope_theta": 500000.0,
"rope_scaling": null,
"sliding_window": null,
"use_sliding_window": false,
"max_window_layers": 28,
"tie_word_embeddings": true,
"torch_dtype": "bfloat16",
"use_cache": true,
"vocab_size": 64000
}