Skip to content

chore: benchmark flags for dataset config and implementation subset #64

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
AmitMY merged 1 commit into from
Jul 24, 2026
Merged

chore: benchmark flags for dataset config and implementation subset #64

Changes from all commits
Commits
Show all changes
1 commit
Select commit
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
8 changes: 7 additions & 1 deletion benchmarks/bench_tokenizers.py
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -60,6 +60,10 @@ def run_case(case, impl, texts_file, num_merges):
parser.add_argument("--samples", type=int, default=2000,
help="dataset rows; the default keeps the whole run around 30s; 0 = full split (~2M words)")
parser.add_argument("--merges", type=int, default=100)
parser.add_argument("--dataset-config", default="wikitext-2-raw-v1",
help="e.g. wikitext-103-raw-v1 for the ~50x corpus")
parser.add_argument("--impls", default=",".join(IMPLS),
help="comma-separated subset of implementations to run")
parser.add_argument("--case", choices=CASES, help=argparse.SUPPRESS) # subprocess mode
parser.add_argument("--impl", choices=IMPLS, help=argparse.SUPPRESS)
parser.add_argument("--texts-file", help=argparse.SUPPRESS)
Expand All @@ -74,11 +78,13 @@ def run_case(case, impl, texts_file, num_merges):
impls = IMPLS
except ImportError:
impls = IMPLS[:-1]
impls = [i for i in impls if i in args.impls.split(",")]

# Load the corpus once and hand it to subprocesses through a file:
# importing datasets and streaming rows costs seconds per subprocess.
from tests.utils import text_dataset
texts = [t for t in text_dataset(max_samples=args.samples or None) if t.strip()]
texts = [t for t in text_dataset(max_samples=args.samples or None,
dataset_config=args.dataset_config) if t.strip()]
with tempfile.NamedTemporaryFile("w", suffix=".txt", delete=False) as f:
f.write("\x00".join(texts))
texts_file = f.name
Expand Down
Loading