Skip to content

ENH: Improve MetaMath training script runtime #2894

New issue

Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.

Sign up for GitHub

By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.

Already on GitHub? Sign in to your account

Merged
merged 2 commits into from
Nov 13, 2025
Merged

ENH: Improve MetaMath training script runtime #2894

Changes from all commits
Commits
Show all changes
2 commits
Select commit Hold shift + click to select a range
3013aa3
ENH Improve MetaMath training script runtime
BenjaminBossan Nov 4, 2025
bc8d9c5
Reviewer feedback: remove obsolete comment
BenjaminBossan Nov 4, 2025
File filter

Filter by extension

Filter by extension
Conversations
Failed to load comments.
Loading
Jump to
Jump to file
Failed to load files.
Loading
Diff view
Diff view
16 changes: 6 additions & 10 deletions method_comparison/MetaMathQA/run.py
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -18,7 +18,6 @@

import argparse
import datetime as dt
import gc
import json
import os
import random
Expand Down Expand Up @@ -58,12 +57,10 @@
from peft.utils import CONFIG_NAME, infer_device


# # suppress all warnings
# warnings.filterwarnings("ignore") # FIXME?

dtype_to_bytes_linear = {"float32": 4, "float16": 2, "bfloat16": 2, "int8": 1, "int4": 0.5}
# if lr scheduler with warmup is used, the ratio of warmup steps to total steps
BUCKET_FACTOR = 20 # number of batches per bucket, increasing this further has diminishing returns
# number of batches per bucket, increasing this further has diminishing returns
BUCKET_FACTOR = 20
# empty device cache every N steps; 10 is a good compromise between keeping memory down while lowering runtime overhead
ACCELERATOR_EMPTY_CACHE_SCHEDULE = 10


def get_generation_config(*, seq_len, generate_kwargs) -> GenerationConfig:
Expand Down Expand Up @@ -298,9 +295,8 @@ def train(
}
print_verbose(json.dumps(log_dict))

# # TODO is this needed?
torch_accelerator_module.empty_cache()
gc.collect()
if step % ACCELERATOR_EMPTY_CACHE_SCHEDULE == 0:
torch_accelerator_module.empty_cache()

print_verbose(f"Training finished after {max_steps} steps, evaluation on test set follows.")
# test set evaluation
Expand Down
Loading