reduce dataset_num_proc to 1/4 of cpu_count

This commit is contained in:
Roland Tannous 2026-02-18 20:53:32 +00:00
commit adc0c78dbc

View file

@ -707,7 +707,7 @@ class UnslothTrainer:
"output_dir": output_dir,
"report_to": ["wandb"] if training_args.get('enable_wandb', False) else "none",
"include_num_input_tokens_seen": True, # Enable token counting
"dataset_num_proc": max(1, os.cpu_count() // 3),
"dataset_num_proc": max(1, os.cpu_count() // 4),
}
# Add warmup parameter - use warmup_ratio if provided, otherwise warmup_steps