Skip to content

don't log batch values if batch is not enabled #2044

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 1 commit into from
Oct 16, 2025
Merged

don't log batch values if batch is not enabled #2044

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
38 changes: 20 additions & 18 deletions gptqmodel/models/base.py
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -735,29 +735,31 @@ def _maybe_resolve_length(value, source_name):
collate_data(sorted_dataset[start: start + batch_size], self.tokenizer.pad_token_id)
for start in range(0, len(sorted_dataset), batch_size)
]
else:
new_calibration_dataset_batched = [
{"input_ids": torch.tensor(block["input_ids"], dtype=torch.long)}
for block in sorted_dataset
]

# total tokens counters
total_padded = 0
total_non_padded = 0
# total tokens counters
total_padded = 0
total_non_padded = 0

for batch in new_calibration_dataset_batched:
# attention_mask is shape [batch_size, seq_len]
mask = batch["attention_mask"]
for batch in new_calibration_dataset_batched:
# attention_mask is shape [batch_size, seq_len]
mask = batch["attention_mask"]

# count where mask == 0 (padded tokens)
total_padded += (mask == 0).sum().item()
# count where mask == 0 (padded tokens)
total_padded += (mask == 0).sum().item()

# count where mask == 1 (non-padded tokens)
total_non_padded += (mask == 1).sum().item()
# count where mask == 1 (non-padded tokens)
total_non_padded += (mask == 1).sum().item()

log.info(f"Calibration: Total padded tokens: {total_padded}")
log.info(f"Calibration: Total non-padded tokens: {total_non_padded}")
log.info(f"Calibration: Total tokens: {total_non_padded + total_padded}")
log.info(f"Calibration: Total padded tokens: {total_padded}")
log.info(f"Calibration: Total non-padded tokens: {total_non_padded}")
log.info(f"Calibration: Total tokens: {total_non_padded + total_padded}")
else:
new_calibration_dataset_batched = [
{
"input_ids": torch.tensor(block["input_ids"], dtype=torch.long),
}
for block in sorted_dataset
]

return new_calibration_dataset_batched

Expand Down