4 Commits
Author SHA1 Message Date
ViperEkura a74e5b91a3 feat: add record-mode to Store for DPO/GRPO
- Store gains fetch_record/num_records alongside stream fetch/__len__
- save_bin/load_bin support per-record offsets via record_keys param
- H5Store/MmapStore/JsonlStore all support dual stream+record access
- DPODataset/GRPODataset use fetch_record, no cross-record concat
- dpo_collate_fn + collate_fn wired through TrainConfig
- fixes attention context leakage in DPO from windowed concatenation
2026-07-18 21:02:29 +08:00
ViperEkura 27524ad085 fix: reset sampler iter at epoch end so progress bar shows total after first epoch 2026-07-04 06:35:55 +08:00
ViperEkura a548d4553e fix: 断点续训恢复优化器/调度器状态及采样器剩余长度
- 使用Checkpoint.load()替代手动加载model.safetensors,恢复optimizer/scheduler状态
- TrainContextBuilder从checkpoint.extra恢复优化器和调度器state_dict
- ResumableDistributedSampler.__len__返回剩余样本数而非总数
- 训练前对state_dict置空避免mp.spawn pickle 7GB大对象
2026-05-26 13:50:25 +08:00
ViperEkura 9c31d78a22 chore: 将data 模块命名为dataset 2026-04-04 16:16:27 +08:00