ViperEkura/AstrAI
1
0
Fork 0
Code Issues Pull Requests Packages Projects Releases Wiki Activity
584 Commits 1 Branch 14 Tags
4de42d83c22a1f5119b83b8d914f7e50646b73d7
Commit Graph
4 Commits
Author SHA1 Message Date
ViperEkura 37a3036934 refactor: split LoRA param init into local vars 2026-07-20 16:23:47 +08:00
ccx1324andClaude Opus 4.7 a5678c9185 fix: create LoRA parameters on base weight device instead of CPU 
When `inject_lora()` replaces Linear layers with LoRALinear after the model
has been moved to CUDA, the new lora_A and lora_B parameters were always
created on CPU, causing a device mismatch error during the forward pass.

Now lora_A and lora_B are created on the same device and dtype as the
parent weight, matching the model's current device.

Co-Authored-By: Claude Opus 4.7 <noreply@anthropic.com>
 2026-07-20 16:11:48 +08:00
ViperEkura 65ab69543b refactor: 统一序列化层，消除分散的 I/O 路径 
- Checkpoint 改为 @dataclass，内聚 save/load 方法
- 提取 save_safetensors/load_safetensors/save_json/load_json 共享工具
- 新增 save_model/load_model_config/load_model_weights 模块函数
- automodel 和 lora 统一委托到 serialization 模块
 2026-05-26 16:44:40 +08:00
ViperEkura a4688021bf feat: 新增LoRA微调模块 
- LoRALinear基于register_parameter托管base weight，state_dict路径不变
- inject_lora/merge_lora/save_lora/load_lora完备封装
- 24个单元测试覆盖注入、合并、存取、边界场景
 2026-05-25 20:15:31 +08:00