Skip to content

save/load peft lora #1358

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
Merged
merged 28 commits into from
Mar 1, 2025
Merged

save/load peft lora #1358

merged 28 commits into from
Mar 1, 2025

Conversation

@Qubitium
Copy link
Collaborator

@Qubitium Qubitium commented Feb 27, 2025 •
edited
Loading

  • Load LoraConfig.r
  • HF Load and use LoraConfig.target_modules
  • GPTQModel Load and use LoraConfig.rank_patterns
  • HF Load and use LoraConfig.rank_patterns
  • dynamic compat (based on dynamic generate target_modules and rank_patterns
  • Save LoraConfig (r, target_module, rank_patterns)
  • CI tests for dynamic and non-dynamic

@Qubitium
save eora to hf format
a2dc904 
Signed-off-by: Qubitium <[email protected]>
@Qubitium Qubitium changed the title save eora to hf format [WIP] save eora to hf format Feb 27, 2025
@Qubitium Qubitium changed the title [WIP] save eora to hf format [WIP] save/load peft lora Feb 27, 2025
Qubitium and others added 12 commits February 27, 2025 09:19
@Qubitium
refractor loading cache into AdapterCache cls
3bfe530 
Signed-off-by: Qubitium <[email protected]>
@Qubitium
add lora rank override code from LoraConfig
d87ad46 
Signed-off-by: Qubitium <[email protected]>
@Qubitium
remove peft dependency
f81b077 
Signed-off-by: Qubitium <[email protected]>
@Qubitium
comment on original HF repo path for test files
e042f70 
Signed-off-by: Qubitium <[email protected]>
@Qubitium
clean up HF download logic
31c8d1f 
Signed-off-by: Qubitium <[email protected]>
@Qubitium
save to PEFT compatible format
b22d094 
Signed-off-by: Qubitium <[email protected]>
@ZX-ModelCloud
add test_quant_and_eora_transformers.py
964e5fe 
Signed-off-by: ZX-ModelCloud <[email protected]>
@ZX-ModelCloud
fix missing task_type in adapter_config.json
ab9519e 
Signed-off-by: ZX-ModelCloud <[email protected]>
@Qubitium
fix regex rule prefix not stripped
7ce0e8d 
Signed-off-by: Qubitium <[email protected]>
@Qubitium
push peft compat changes
45935ff 
Signed-off-by: Qubitium <[email protected]>
@Qubitium
prevent preft doing alpha / r scaling. set alpha eq r so math is just…
5ed4472 
… 1, no scaling

Signed-off-by: Qubitium <[email protected]>
@ZX-ModelCloud
fix lora load with transformers
32a468d 
Signed-off-by: ZX-ModelCloud <[email protected]>
@Qubitium Qubitium changed the title [WIP] save/load peft lora save/load peft lora Feb 28, 2025
@Qubitium Qubitium marked this pull request as ready for review February 28, 2025 05:42
ZX-ModelCloud and others added 2 commits February 28, 2025 05:46
@ZX-ModelCloud
fix device
e590454 
Signed-off-by: ZX-ModelCloud <[email protected]>
@Qubitium
format
95f3e15 
Signed-off-by: Qubitium <[email protected]>
@Qubitium Qubitium mentioned this pull request Feb 28, 2025
Merged
1 task
@Qubitium Qubitium merged commit f7b86a5 into main Mar 1, 2025
4 checks passed
@Qubitium Qubitium deleted the lora-format branch March 4, 2025 12:17
Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment

Reviewers

No reviews

Assignees

No one assigned

Labels

None yet

Projects

None yet

Milestone

No milestone

Development

Successfully merging this pull request may close these issues.

4 participants

@Qubitium @ZX-ModelCloud @CSY-ModelCloud