-
Notifications
You must be signed in to change notification settings - Fork 688
Export a lora model #11045
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
Export a lora model #11045
Conversation
^ Program+data combined currently, using the lora linear definition. Differential Revision: [D75153377](https://our.internmc.facebook.com/intern/diff/D75153377/) [ghstack-poisoned]
🔗 Helpful Links🧪 See artifacts and rendered test results at hud.pytorch.org/pr/pytorch/executorch/11045
Note: Links to docs will display an error until the docs builds have been completed. ❌ 2 New Failures, 2 Unrelated FailuresAs of commit bda599b with merge base 8da2ea6 ( NEW FAILURES - The following jobs have failed:
BROKEN TRUNK - The following jobs failed but were present on the merge base:👉 Rebase onto the `viable/strict` branch to avoid these failures
This comment was automatically generated by Dr. CI and updates every 15 minutes. |
^ Program+data combined currently, using the lora linear definition. Differential Revision: [D75153377](https://our.internmc.facebook.com/intern/diff/D75153377/) ghstack-source-id: 285402970 Pull Request resolved: #11045
This pull request was exported from Phabricator. Differential Revision: D75153377 |
adapter_checkpoint = {} | ||
adapter_config = {} | ||
adapter_checkpoint_path = kwargs.get("adapter_checkpoint", None) | ||
if adapter_checkpoint_path: |
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.
nit: use os.exists or similar from Path?
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.
tests?
^ Program+data combined currently, using the lora linear definition. Differential Revision: [D75153377](https://our.internmc.facebook.com/intern/diff/D75153377/) [ghstack-poisoned]
Pull Request resolved: #11045 ^ Program+data combined currently, using the lora linear definition. Differential Revision: [D75153377](https://our.internmc.facebook.com/intern/diff/D75153377/) ghstack-source-id: 295432560
Let me add one. Maybe with phi-3-mini or a smaller model. |
metadata: Json string containing metadata information. | ||
e.g. '"{\"get_bos_id\":128000, \"get_eos_ids\":[128009, 128001]}"' | ||
use_lora: Rank of the LoRA, if set to 0 then this means no LoRA. For use with QAT. | ||
use_lora: Only for use with QAT. Rank of the LoRA adapter, disabled |
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.
You didnt add this, but why is this boolean named field an int, and why does it correspond with qat?
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.
cc @cccclai ?
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.
Didn't add it myself either, I think it's likely from Lunwen - I believe it's for the llama 3.2 1b QAT checkpoint which include LoRA, so make sure we don't break llama3.2 QAT model if we use this flag somewhere else
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.
^ Program+data combined currently, using the lora linear definition. Issue: #9580 Differential Revision: [D75153377](https://our.internmc.facebook.com/intern/diff/D75153377/) [ghstack-poisoned]
Pull Request resolved: #11045 ^ Program+data combined currently, using the lora linear definition. Differential Revision: [D75153377](https://our.internmc.facebook.com/intern/diff/D75153377/) ghstack-source-id: 295432560
This pull request was exported from Phabricator. Differential Revision: D75153377 |
^ Program+data combined currently, using the lora linear definition. Issue: #9580 Differential Revision: [D75153377](https://our.internmc.facebook.com/intern/diff/D75153377/) [ghstack-poisoned]
This pull request was exported from Phabricator. Differential Revision: D75153377 |
^ Program+data combined currently, using the lora linear definition. Issue: #9580 Differential Revision: [D75153377](https://our.internmc.facebook.com/intern/diff/D75153377/) [ghstack-poisoned]
^ Program+data combined currently, using the lora linear definition. Issue: #9580 Differential Revision: [D75153377](https://our.internmc.facebook.com/intern/diff/D75153377/) [ghstack-poisoned]
^ Program+data combined currently, using the lora linear definition. Issue: #9580 Differential Revision: [D75153377](https://our.internmc.facebook.com/intern/diff/D75153377/) [ghstack-poisoned]
^ Program+data combined currently, using the lora linear definition. Issue: #9580 Differential Revision: [D75153377](https://our.internmc.facebook.com/intern/diff/D75153377/) [ghstack-poisoned]
^ Program+data combined currently, using the lora linear definition. Issue: #9580 Differential Revision: [D75153377](https://our.internmc.facebook.com/intern/diff/D75153377/) [ghstack-poisoned]
^ Program+data combined currently, using the lora linear definition. Issue: #9580 Differential Revision: [D75153377](https://our.internmc.facebook.com/intern/diff/D75153377/) [ghstack-poisoned]
Pull Request resolved: #11045 ^ Program+data combined currently, using the lora linear definition. ghstack-source-id: 298641176 Differential Revision: [D75153377](https://our.internmc.facebook.com/intern/diff/D75153377/)
This pull request was exported from Phabricator. Differential Revision: D75153377 |
9558550
into
gh/lucylq/83/base
Stack from ghstack (oldest at bottom):
^
Program+data combined currently, using the lora linear definition.
Issue: #9580
Differential Revision: D75153377