mirror of
https://github.com/osmarks/nanogpt-experiments.git
synced 2024-12-21 15:40:28 +00:00
26 lines
677 B
Python
26 lines
677 B
Python
# config for training GPT-2 (124M) down to very nice loss of ~2.85 on 1 node of 8X A100 40GB
|
|
# launch as the following (e.g. in a screen session) and wait ~5 days:
|
|
# $ torchrun --standalone --nproc_per_node=8 train.py config/train_gpt2.py
|
|
|
|
wandb_log = True
|
|
wandb_project = 'owt'
|
|
wandb_run_name='gpt2-124M'
|
|
|
|
# these make the total batch size be ~0.5M
|
|
# 12 batch size * 1024 block size * 5 gradaccum * 8 GPUs = 491,520
|
|
batch_size = 12
|
|
block_size = 1024
|
|
gradient_accumulation_steps = 5
|
|
|
|
# this makes total number of tokens be 300B
|
|
max_iters = 600000
|
|
lr_decay_iters = 600000
|
|
|
|
# eval stuff
|
|
eval_interval = 1000
|
|
eval_iters = 200
|
|
log_interval = 10
|
|
|
|
# weight decay
|
|
weight_decay = 1e-1
|