tacit/nanochat
1
0
Fork 0
mirror of https://github.com/karpathy/nanochat.git synced 2026-04-01 21:25:21 +00:00
Code Issues Actions Packages Projects Releases Wiki Activity

Merge 9f9ef95adc into 7808dc7159

Browse Source
This commit is contained in:
Suraj-Self 2026-03-26 08:29:32 +05:30 committed by GitHub
commit 8db278572a
No known key found for this signature in database
GPG Key ID: B5690EEEBB952194
2 changed files with 2 additions and 2 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

2
scripts/base_train.py
View File

@ -406,7 +406,7 @@ else:
# Figure out the needed gradient accumulation micro-steps to reach the desired total batch size per step
tokens_per_fwdbwd = args.device_batch_size * args.max_seq_len # tokens per iteration for a single rank
world_tokens_per_fwdbwd = tokens_per_fwdbwd * ddp_world_size # total tokens per iteration for all ranks
assert total_batch_size % world_tokens_per_fwdbwd == 0
assert total_batch_size % world_tokens_per_fwdbwd == 0, f"total_batch_size ({total_batch_size}) must be a multiple of {world_tokens_per_fwdbwd}."
grad_accum_steps = total_batch_size // world_tokens_per_fwdbwd
print0(f"Tokens / micro-batch / rank: {args.device_batch_size} x {args.max_seq_len} = {tokens_per_fwdbwd:,}")
print0(f"Tokens / micro-batch: {world_tokens_per_fwdbwd:,}")

2
scripts/chat_sft.py
View File

@ -122,7 +122,7 @@ depth = model.config.n_layer
num_flops_per_token = model.estimate_flops()
tokens_per_fwdbwd = args.device_batch_size * args.max_seq_len # tokens per iteration for a single rank
world_tokens_per_fwdbwd = tokens_per_fwdbwd * ddp_world_size # total tokens per iteration for all ranks
assert args.total_batch_size % world_tokens_per_fwdbwd == 0
assert args.total_batch_size % world_tokens_per_fwdbwd == 0, f"total_batch_size ({args.total_batch_size}) must be a multiple of {world_tokens_per_fwdbwd}."
grad_accum_steps = args.total_batch_size // world_tokens_per_fwdbwd
print0(f"Tokens / micro-batch / rank: {args.device_batch_size} x {args.max_seq_len} = {tokens_per_fwdbwd:,}")
print0(f"Tokens / micro-batch: {world_tokens_per_fwdbwd:,}")