tocmo0nlord/axolotl
1
0
Fork 0
Code Issues Pull Requests Actions 2 Packages Projects Releases Wiki Activity

Default micro_batch_size to 1 for a safer start

Browse Source
This commit is contained in:
Utensil
2023-06-03 17:26:33 +08:00
committed by GitHub
parent ca11ae9689
commit c9c050316f
1 changed files with 1 additions and 1 deletions
Download Patch File Download Diff File Expand all files Collapse all files

2
examples/falcon/config-7b-qlora.yml
View File

@@ -50,7 +50,7 @@ output_dir: ./qlora-out
# - 7b: 40 # - 7b: 40
# - 40b: 4 # - 40b: 4
# decrease if OOM, increase for max VRAM utilization # decrease if OOM, increase for max VRAM utilization
micro_batch_size: 30 micro_batch_size: 1
gradient_accumulation_steps: 2 gradient_accumulation_steps: 2
num_epochs: 3 num_epochs: 3
# Optimizer for QLoRA # Optimizer for QLoRA