Skip to content

fix(minillm): qwen2 teacher student model unmatched model vocab size #331

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

Open
anjilab wants to merge 1 commit into
base: main
Choose a base branch
from
Open

fix(minillm): qwen2 teacher student model unmatched model vocab size #331

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
4 changes: 4 additions & 0 deletions minillm/finetune.py
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -159,6 +159,10 @@ def get_distil_loss(args, tokenizer, model, teacher_model, model_batch, no_model
teacher_model.eval()
teacher_outputs = teacher_model(**model_batch, use_cache=False)
teacher_logits = teacher_outputs.logits
if args.model_type == 'qwen2':
# If ZeRO, Get vocab size under module, No ZeRO - directly from config
student_vocab_size = model.module.config.vocab_size if hasattr(model, "module") else model.config.vocab_size
teacher_logits = teacher_logits[:, :, :student_vocab_size]
if args.model_parallel:
distil_losses = mpu.parallel_soft_cross_entropy_loss(logits.float(), teacher_logits.float())
distil_losses = distil_losses.view(-1)
Expand Down