Skip to content
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

Jump to bottom

fix llama2 70b lora tuning bug #7622

Merged
merged 3 commits into from
Oct 4, 2023
Merged

fix llama2 70b lora tuning bug #7622

Changes from all commits
Commits
Show all changes
3 commits
Select commit Hold shift + click to select a range
8197b83
fix llama2 70b lora tuning bug
cuichenx Oct 4, 2023
ef9f8c1
Update peft_config.py
arendu Oct 4, 2023
b1908ee
Merge branch 'main' into chcui/peft_refactor_llama2_70b
cuichenx Oct 4, 2023
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
6 changes: 5 additions & 1 deletion nemo/collections/nlp/parts/peft_config.py
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -60,10 +60,14 @@ def __init__(self, cfg):
else:
kv_channels = cfg.kv_channels
projection_size = kv_channels * cfg.num_attention_heads
num_query_groups = cfg.get("num_query_groups", None)
if num_query_groups is None:
num_query_groups = cfg.num_attention_heads
qkv_projection_size = projection_size + (2 * kv_channels * num_query_groups)
Copy link
Collaborator

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

will this continue to work for non-group query attention models?

Copy link
Collaborator Author

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

Yes! Tested with llama2 7b as well.


config_args = {
"in_features": cfg.hidden_size,
"out_features": 3 * projection_size,
"out_features": qkv_projection_size,
"dim": lora_cfg.adapter_dim,
"norm_position": None,
"norm_type": None,
Expand Down
Loading