songsenand
  • Joined on 2024-01-17
songsenand pushed to main at songsenand/SUInput 2026-02-22 10:20:38 +08:00
96706abb93 feat(model): 修改池化层输出维度并添加线性变换层
songsenand pushed to main at songsenand/SUInput 2026-02-22 09:31:07 +08:00
fc71124484 feat(suinput): 引入拼音分组配置并优化上下文采样逻辑
songsenand pushed to main at songsenand/SUInput 2026-02-21 22:03:08 +08:00
2219f6530d 更新评估数据集样本文件
songsenand pushed to main at songsenand/SUInput 2026-02-21 22:01:45 +08:00
51f9ddbc70 修复拼音组处理逻辑,避免未处理拼音导致的索引错误
songsenand pushed to main at songsenand/SUInput 2026-02-21 21:56:05 +08:00
8f58917d13 调整拼音分组与采样逻辑,优化模型结构及专家路由策略
songsenand pushed to main at songsenand/SUInput 2026-02-21 00:56:23 +08:00
917c9f4256 调整数据采样逻辑以提升模型训练效果
songsenand pushed to main at songsenand/SUInput 2026-02-20 23:30:44 +08:00
17324ffa10 修复初始步骤损失计算逻辑
songsenand pushed to main at songsenand/SUInput 2026-02-20 23:28:47 +08:00
4560a9ed06 移除 global_step 自增逻辑并调整至循环末尾
songsenand pushed to main at songsenand/SUInput 2026-02-20 23:21:46 +08:00
558d7f9fc9 调整模型结构及参数以优化性能
songsenand pushed to main at songsenand/SUInput 2026-02-16 10:27:01 +08:00
ae414bae6b feat(trainer): 添加残差块以增强模型表达能力
songsenand pushed to main at songsenand/SUInput 2026-02-15 23:01:58 +08:00
ab2dbc378b 修复损失权重计算逻辑,修正平方根次数以提升稳定性
songsenand pushed to main at songsenand/SUInput 2026-02-15 21:51:39 +08:00
cd25349d90 删除旧的 MoE 模型文件
songsenand pushed to main at songsenand/SUInput 2026-02-15 01:48:50 +08:00
0d529c0c89 调整损失权重计算并优化训练循环终止条件
songsenand pushed to main at songsenand/SUInput 2026-02-15 01:07:06 +08:00
94b44e6f71 添加损失权重支持并重构部分模块结构
songsenand pushed to main at songsenand/SUInput 2026-02-15 00:25:48 +08:00
515f261824 修复模型加载方法,使用正确的实例方法加载状态字典
songsenand pushed to main at songsenand/SUInput 2026-02-15 00:08:59 +08:00
fd913748ca 调整残差块和分类头的 dropout 概率,并新增残差模块到 MoE 模型
songsenand pushed to main at songsenand/SUInput 2026-02-14 23:34:41 +08:00
e91f823d65 feat: 优化模型输入处理与专家数量,增强训练与推理兼容性
songsenand pushed to main at songsenand/SUInput 2026-02-14 17:07:42 +08:00
9fad2bf1d4 修复损失计算方式,使用NLLLoss替代原始criterion
songsenand pushed to main at songsenand/SUInput 2026-02-14 15:51:11 +08:00
f89635b201 添加 package-data 配置以包含 trainer 和 suinput 模块的额外数据文件
songsenand pushed to main at songsenand/SUInput 2026-02-14 15:43:31 +08:00
d60997438e 更新评估数据集样本文件