xinyun/vllm
1
0
Fork 0
mirror of https://git.datalinker.icu/vllm-project/vllm.git synced 2026-06-03 20:22:18 +08:00
Code Issues Packages Projects Releases Wiki Activity

[Bugfix] LoRA : Fix the order in which the kernels process LoRAs (#16040)

Browse Source 
Signed-off-by: Varun Sundar Rabindranath <varun@neuralmagic.com>
Co-authored-by: Varun Sundar Rabindranath <varun@neuralmagic.com>
This commit is contained in:
Varun Sundar Rabindranath 2025-04-06 10:04:50 -04:00 committed by GitHub
parent 242a637aea
commit 3a100b9278
No known key found for this signature in database
GPG Key ID: B5690EEEBB952194
1 changed files with 1 additions and 1 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

2
vllm/lora/ops/triton_ops/lora_kernel_metadata.py
View File

@ -111,7 +111,7 @@ class LoRAKernelMeta:
# active_lora_ids, num_tokens_per_lora # active_lora_ids, num_tokens_per_lora
lora_ids, num_tokens_per_lora = torch.unique(token_lora_mapping, lora_ids, num_tokens_per_lora = torch.unique(token_lora_mapping,
sorted=False, sorted=True,
return_counts=True) return_counts=True)
self.active_lora_ids[:lora_ids.size(0)].copy_(lora_ids, self.active_lora_ids[:lora_ids.size(0)].copy_(lora_ids,
non_blocking=True) non_blocking=True)