Files
Test_context_pretrain/config.json
ModelHub XC 200675bd2d 初始化项目,由ModelHub XC社区提供模型
Model: MathMindsAGI/Test_context_pretrain
Source: Original Platform
2026-04-11 11:04:57 +08:00

30 lines
691 B
JSON

{
"architectures": [
"Qwen2ForCausalLM"
],
"attention_dropout": 0.0,
"bos_token_id": 2,
"eos_token_id": 3,
"hidden_act": "silu",
"hidden_size": 768,
"initializer_range": 0.02,
"intermediate_size": 3072,
"max_position_embeddings": 2048,
"max_window_layers": 24,
"model_type": "qwen2",
"num_attention_heads": 12,
"num_hidden_layers": 12,
"num_key_value_heads": 2,
"rms_norm_eps": 1e-06,
"rope_scaling": null,
"rope_theta": 1000000.0,
"sliding_window": 2048,
"tie_word_embeddings": true,
"torch_dtype": "float32",
"transformers_version": "4.52.4",
"use_cache": false,
"use_mrope": false,
"use_sliding_window": false,
"vocab_size": 2200
}