Open NathanHB opened 2 months ago
lighteval.commands.parsers
how to use:
nothing changes except for the use of the cli executable
lighteval nanotron {nanotron args}
accelerate args now need to be passed to the lighteval script.
accelerate launch --multi_gpu --num_processes=8 -m \ lighteval accelerate \ --model_args "pretrained=gpt2,dtype=float16" \ --tasks "original|mmlu:anatomy|0|0" \ --override_batch_size 1 \ --output_dir="tmp/" \ --max_samples 1 \ --use_chat_template
lighteval.commands.parsers
how to use:
with nanotron
nothing changes except for the use of the cli executable
with accelerate
accelerate args now need to be passed to the lighteval script.