admin/lerobot
1
0
Fork 0
mirror of https://github.com/huggingface/lerobot.git synced 2026-05-18 02:00:03 +00:00
Code Issues Packages Projects Releases Wiki Activity

fix formatting

Browse Source
This commit is contained in:
Pepijn
2025-10-14 17:37:47 +02:00
parent f8a185f753
commit ebf64bd80e
6 changed files with 50 additions and 65 deletions
Download Patch File Download Diff File Expand all files Collapse all files
docs/source/multi_gpu_training.mdx
+3 -3
View File
@@ -80,7 +80,7 @@ When you launch training with accelerate:
### Why No Automatic Scaling?
Many distributed training frameworks automatically scale the learning rate by the number of GPUs (e.g., `lr = base_lr × num_gpus`).
Many distributed training frameworks automatically scale the learning rate by the number of GPUs (e.g., `lr = base_lr × num_gpus`).
However, LeRobot keeps the learning rate exactly as you specify it.
### When and How to Scale
@@ -104,8 +104,8 @@ Since the effective batch size `bs` increases with multiple GPUs (batch_size ×
```bash
# Example: 2 GPUs with effective batch size 2x larger
# Original: batch_size=8, steps=100000
# With 2 GPUs: batch_size=8 (16 in total), steps=50000
# Original: batch_size=8, steps=100000
# With 2 GPUs: batch_size=8 (16 in total), steps=50000
accelerate launch --num_processes=2 $(which lerobot-train) \
--batch_size=8 \
--steps=50000 \