diff options
| author | YurenHao0426 <Blackhao0426@gmail.com> | 2026-04-13 00:22:32 -0500 |
|---|---|---|
| committer | YurenHao0426 <Blackhao0426@gmail.com> | 2026-04-13 00:22:32 -0500 |
| commit | 3b2b49845a256fcabc55af789562ca034bb69ebe (patch) | |
| tree | 6e3d82e8f4e248963509d091b31a9bf18f6517cc /baselines/prompt_all_k.py | |
| parent | bfdcc36c0e31adfa95410ce87e7da646e0b948fe (diff) | |
Fix VeRA crash: reload model fresh before each PEFT method
Root cause: get_peft_model() modifies model in-place. After LoRA/TinyLoRA
cleanup, the model's modules are altered so VeRA can't find target_modules.
Fix: reload AutoModelForCausalLM from scratch before each PEFT method.
Slower but reliable — no more cross-contamination between PEFT methods.
Co-Authored-By: Claude Opus 4.6 (1M context) <noreply@anthropic.com>
Diffstat (limited to 'baselines/prompt_all_k.py')
0 files changed, 0 insertions, 0 deletions