LoRA support in vllm and hf backends#56

Merged

vicky-xef merged 51 commits intomainfrom

Jan 30, 2026

Commits on Nov 20, 2025

Add lora in vllm & some tests
vicky-xef
committed

Commits on Nov 21, 2025

add batched method in async + more tests
vicky-xef
committed
decrease difference error for lora because of precision issues (e.g. bfloat16)
vicky-xef
committed

Commits on Nov 24, 2025

set lora_request as class attribute
vicky-xef
committed

Commits on Nov 25, 2025

change hf backend to support lora + add testing
vicky-xef
committed
clean hf lora tests
vicky-xef
committed
add testing for swapping lora and no-lora
vicky-xef
committed
remove unnecessary import
vicky-xef
committed
remove double batch method
vicky-xef
committed

Commits on Dec 3, 2025

add comments in the new methods
vicky-xef
committed
remove comment
vicky-xef
committed

Commits on Dec 5, 2025

add more tests
vicky-xef
committed

Commits on Dec 8, 2025

update dependencies
vicky-xef
committed
cleaning
vicky-xef
committed

Commits on Dec 9, 2025

change model for testing
vicky-xef
committed
add lora dependencies in pytest
vicky-xef
committed
fix dependencies lora
vicky-xef
committed
change test model
vicky-xef
committed

Commits on Dec 10, 2025

fix lora test on transformer
vicky-xef
committed
increase gpu memory util
vicky-xef
committed
decrease gpu memory util
vicky-xef
committed

Commits on Dec 11, 2025

check gpu github
vicky-xef
committed
change gpu memory util
vicky-xef
committed
debug github
vicky-xef
committed
decrease tests
vicky-xef
committed

Commits on Dec 12, 2025

downgrade triton
vicky-xef
committed
trition 3.2
vicky-xef
committed
debug models github
vicky-xef
committed
change model on tests
vicky-xef
committed
remove test for cache reasons
vicky-xef
committed
free disk space
vicky-xef
committed
triton
vicky-xef
committed
add testing for error path
vicky-xef
committed
add readme
vicky-xef
committed
cleaning
vicky-xef
committed

Commits on Dec 15, 2025

triton reinstall
vicky-xef
committed
triton 3.2
vicky-xef
committed
remove unnecessary reinstall
vicky-xef
committed

Commits on Dec 16, 2025

adding lora methods in base class
vicky-xef
committed
no cover
vicky-xef
committed
change triton version
vicky-xef
committed

Commits on Dec 17, 2025

trition uninstall and install
vicky-xef
committed
rm triton caches
vicky-xef
committed
dependencies explicitly
vicky-xef
committed
dependencies explicitly
vicky-xef
committed
triton reinstall
vicky-xef
committed

Commits on Dec 19, 2025

rename methods, add add_new_lora to vllm
vicky-xef
committed

Commits on Jan 12, 2026

deps fix
vicky-xef
committed
Merge branch 'main' into vicky/lora
vicky-xef
authored
attributes fix
vicky-xef
committed
fix tests
vicky-xef
committed