Reduce memory usage
a96d072
-
SmolLM2-135M-Instruct-TaiwanChat
Sync train log to W&B service
-
runs
Model save
-
1.52 kB
initial commit
-
1.71 kB
Model save
-
879 Bytes
Model save
-
132 Bytes
Training in progress, step 5000
-
466 kB
Training in progress, step 5000
-
269 MB
Model save
-
655 Bytes
Training in progress, step 5000
-
3.52 MB
Model save
-
3.79 kB
Training in progress, step 5000
-
5.56 kB
Reduce memory usage
-
6.71 kB
rename wandb project name for train with unsloth .py
training_args.bin
Detected Pickle imports (10)
- "transformers.trainer_pt_utils.AcceleratorConfig",
- "transformers.trainer_utils.IntervalStrategy",
- "accelerate.state.PartialState",
- "torch.device",
- "transformers.training_args.TrainingArguments",
- "accelerate.utils.dataclasses.DistributedType",
- "transformers.training_args.OptimizerNames",
- "transformers.trainer_utils.HubStrategy",
- "transformers.trainer_utils.SchedulerType",
- "transformers.trainer_utils.SaveStrategy"
How to fix it?
5.37 kB
Model save
-
801 kB
Training in progress, step 5000
