-
Notifications
You must be signed in to change notification settings - Fork 3.2k
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.
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
Use LLaMA impl of Huggingface Transformers #2263
Conversation
There was a problem hiding this comment.
Choose a reason for hiding this comment
The reason will be displayed to describe this comment to others. Learn more.
LGTM
source_url: https://huggingface.co/datasets/laion/OIG/resolve/main/unified_chip2.jsonl | ||
max_count: 10000 | ||
min_length: 1000 | ||
val_split: 0.2 |
There was a problem hiding this comment.
Choose a reason for hiding this comment
The reason will be displayed to describe this comment to others. Learn more.
for 10k samples, val_split=0.1 is probably enough
There was a problem hiding this comment.
Choose a reason for hiding this comment
The reason will be displayed to describe this comment to others. Learn more.
yes definitely, now working on the effective values in other pr.
@@ -98,7 +98,7 @@ oasst_export_eu: | |||
- alpaca | |||
- oig_file: | |||
source_url: https://huggingface.co/datasets/laion/OIG/resolve/main/unified_chip2.jsonl | |||
max_count: 30000 | |||
max_count: 10000 | |||
min_length: 1000 | |||
val_split: 0.2 |
There was a problem hiding this comment.
Choose a reason for hiding this comment
The reason will be displayed to describe this comment to others. Learn more.
Same here, val_split 0.1 is probably enough
There was a problem hiding this comment.
Choose a reason for hiding this comment
The reason will be displayed to describe this comment to others. Learn more.
When pre-training works I will probably remove this here .. the GPT4all dataset re-generade all chip answers with GPT3.5 .. we'll use them.
No description provided.