[Performance] Speed up model loading by desynchronizing sync ops in `from_accelerate`
March 23, 2026 · #647
Python
Difficulty: Medium
Labels
good first issue
Parent Repository
vllm-project/compressed-tensors
Python repository
270 80