fix: 修复 CLI 参数缺失/重复、device_ids 越界、generate 参数名不一致、scheduler 时序、非流式截断等 bug

- train.py: 补上 --batch_size、--grpo_clip_eps,删除 3 处重复 --group_size
- generate.py: --model_dir 改为 --param_path 对齐 README
- automodel.py: from_pretrained 新增 strict 参数(默认 True)
- parallel/setup.py: 修复 device_ids 索引越界
- train_callback.py: scheduler.step() 移至 on_step_end
- test_train_strategy.py: 测试中补 optimizer.step()
- engine.py: 非流式改为循环等待所有任务完成,补 remove_task 清理
- scheduler.py: Task 添加 _pages_freed 标志,杜绝双重释放
- trainer.py: accumulation_steps=0 时 clamp 为 1
- tokenizer.py: save_pretrained 添加 _tokenizer is None 检查
- benchmark.py: 修复 ModelConfig 过时 import 路径
- inference/__init__.py: 修复 stale docstring
This commit is contained in:
2026-05-09 14:36:42 +08:00
parent bc7c82977e
commit 283bcaf2ff
12 changed files with 49 additions and 30 deletions
+3 -2
View File
@@ -48,8 +48,9 @@ def setup_parallel(
if device_ids is None:
device_ids = [i for i in range(world_size)]
rank = device_ids[rank % len(device_ids)]
device_id = torch.device(device_type, device_ids[rank])
effective_rank = rank % len(device_ids)
device_id = torch.device(device_type, device_ids[effective_rank])
rank = device_ids[effective_rank]
os.environ["MASTER_ADDR"] = master_addr
os.environ["MASTER_PORT"] = master_port