admin/lerobot
1
0
Fork 0
mirror of https://github.com/huggingface/lerobot.git synced 2026-05-23 12:40:08 +00:00
Code Issues Packages Projects Releases Wiki Activity

fix(config): add lora_alpha to PeftConfig (#3573)

Browse Source 
* fix(config): add lora_alpha to PeftConfig

PeftConfig was missing the lora_alpha field, causing the PEFT library
to default to alpha=8 regardless of the LoRA rank, which dampens the
adaptation signal for high-rank adapters (e.g., r=128).

This adds lora_alpha: int | None = None to PeftConfig, allowing users
to specify --peft.lora_alpha <value> on the CLI.

Closes #3551

* fix(docs): add lora_alpha to peft training example + clarify scaling formula

- Add --peft.lora_alpha=64 to docs/source/peft_training.mdx example to
  prevent new users from hitting the alpha=8 default dampening bug
- Clarify lora_alpha comment in default.py with scaling = lora_alpha / r

* docs: mention both --peft.r and --peft.lora_alpha in LoRA description

---------

Co-authored-by: Cheng Yin <yin@users.noreply.github.com>
This commit is contained in:
Cheng Yin
2026-05-13 17:09:19 +08:00
committed by GitHub
parent fe96b28c74
commit 9db9c35cb4
2 changed files with 10 additions and 2 deletions
Download Patch File Download Diff File Expand all files Collapse all files
docs/source/peft_training.mdx
+4 -2
View File
@@ -28,13 +28,15 @@ lerobot-train \
--steps=100000 \ --steps=100000 \
--batch_size=32 \ --batch_size=32 \
--peft.method_type=LORA \ --peft.method_type=LORA \
--peft.r=64 --peft.r=64 \
--peft.lora_alpha=64
``` ```
Note the `--peft.method_type` parameter that let's you select which PEFT method to use. Here we use Note the `--peft.method_type` parameter that let's you select which PEFT method to use. Here we use
[LoRA](https://huggingface.co/docs/peft/main/en/package_reference/lora) (Low-Rank Adapter) which is probably the most [LoRA](https://huggingface.co/docs/peft/main/en/package_reference/lora) (Low-Rank Adapter) which is probably the most
popular fine-tuning method to date. Low-rank adaption means that we only fine-tune a matrix with comparably low rank popular fine-tuning method to date. Low-rank adaption means that we only fine-tune a matrix with comparably low rank
instead of the full weight matrix. This rank can be specified using the `--peft.r` parameter. The higher the rank instead of the full weight matrix. This rank can be specified using the `--peft.r` parameter, and the LoRA scaling factor with
`--peft.lora_alpha` (where `scaling = lora_alpha / r`). The higher the rank
the closer you get to full fine-tuning the closer you get to full fine-tuning
There are more complex methods that have more parameters. These are not yet supported, feel free to raise an issue There are more complex methods that have more parameters. These are not yet supported, feel free to raise an issue
src/lerobot/configs/default.py
+6
View File
@@ -117,3 +117,9 @@ class PeftConfig:
# the rank used for the adapter. In general a higher rank means more trainable parameters and closer to full # the rank used for the adapter. In general a higher rank means more trainable parameters and closer to full
# fine-tuning. # fine-tuning.
r: int = 16 r: int = 16
# Alpha parameter for LoRA scaling (scaling = lora_alpha / r).
# In general, a higher alpha means stronger adaptation signal.
# If None, the PEFT library defaults to alpha=8, which may dampen high-rank adapters.
# Common values are r (alpha == rank) or 2*r.
lora_alpha: int | None = None