[Bug]: UnboundLocalError: local variable 'lora_b_k' referenced before assignment

[Stealthwriter]

Your current environment

The output of `python collect_env.py`

Collecting environment information...
PyTorch version: 2.3.0+cu121
Is debug build: False
CUDA used to build PyTorch: 12.1
ROCM used to build PyTorch: N/A

OS: Ubuntu 22.04.3 LTS (x86_64)
GCC version: (Ubuntu 11.4.0-1ubuntu1~22.04) 11.4.0
Clang version: Could not collect
CMake version: version 3.29.3
Libc version: glibc-2.35

Python version: 3.10.12 (main, Nov 20 2023, 15:14:05) [GCC 11.4.0] (64-bit runtime)
Python platform: Linux-5.15.0-75-generic-x86_64-with-glibc2.35
Is CUDA available: True
CUDA runtime version: 12.1.105
CUDA_MODULE_LOADING set to: LAZY
GPU models and configuration:
GPU 0: NVIDIA A100 80GB PCIe
GPU 1: NVIDIA A100 80GB PCIe

Nvidia driver version: 545.23.08
cuDNN version: Could not collect
HIP runtime version: N/A
MIOpen runtime version: N/A
Is XNNPACK available: True

CPU:
Architecture: x86_64
CPU op-mode(s): 32-bit, 64-bit
Address sizes: 48 bits physical, 48 bits virtual
Byte Order: Little Endian
CPU(s): 252
On-line CPU(s) list: 0-251
Vendor ID: AuthenticAMD
Model name: AMD EPYC 7763 64-Core Processor
CPU family: 25
Model: 1
Thread(s) per core: 1
Core(s) per socket: 126
Socket(s): 2
Stepping: 1
BogoMIPS: 4890.81
Flags: fpu vme de pse tsc msr pae mce cx8 apic sep mtrr pge mca cmov pat pse36 clflush mmx fxsr sse sse2 ht syscall nx mmxext fxsr_opt pdpe1gb rdtscp lm rep_good nopl cpuid extd_apicid tsc_known_freq pni pclmulqdq ssse3 fma cx16 pcid sse4_1 sse4_2 x2apic movbe popcnt tsc_deadline_timer aes xsave avx f16c rdrand hypervisor lahf_lm cmp_legacy svm cr8_legacy abm sse4a misalignsse 3dnowprefetch osvw perfctr_core invpcid_single ssbd ibrs ibpb stibp vmmcall fsgsbase tsc_adjust bmi1 avx2 smep bmi2 erms invpcid rdseed adx smap clflushopt clwb sha_ni xsaveopt xsavec xgetbv1 xsaves clzero xsaveerptr wbnoinvd arat npt lbrv nrip_save tsc_scale vmcb_clean pausefilter pfthreshold v_vmsave_vmload vgif umip pku ospke vaes vpclmulqdq rdpid fsrm arch_capabilities
Virtualization: AMD-V
Hypervisor vendor: KVM
Virtualization type: full
L1d cache: 15.8 MiB (252 instances)
L1i cache: 15.8 MiB (252 instances)
L2 cache: 126 MiB (252 instances)
L3 cache: 3.9 GiB (252 instances)
NUMA node(s): 1
NUMA node0 CPU(s): 0-251
Vulnerability Itlb multihit: Not affected
Vulnerability L1tf: Not affected
Vulnerability Mds: Not affected
Vulnerability Meltdown: Not affected
Vulnerability Mmio stale data: Not affected
Vulnerability Retbleed: Not affected
Vulnerability Spec store bypass: Mitigation; Speculative Store Bypass disabled via prctl and seccomp
Vulnerability Spectre v1: Mitigation; usercopy/swapgs barriers and __user pointer sanitization
Vulnerability Spectre v2: Mitigation; Retpolines, IBPB conditional, IBRS_FW, STIBP disabled, RSB filling, PBRSB-eIBRS Not affected
Vulnerability Srbds: Not affected
Vulnerability Tsx async abort: Not affected

Versions of relevant libraries:
[pip3] numpy==1.26.3
[pip3] nvidia-nccl-cu12==2.20.5
[pip3] torch==2.3.0
[pip3] torchaudio==2.2.0
[pip3] torchvision==0.17.0
[pip3] triton==2.3.0
[pip3] vllm_nccl_cu12==2.18.1.0.4.0
[conda] Could not collectROCM Version: Could not collect
Neuron SDK Version: N/A
vLLM Version: 0.4.2
vLLM Build Flags:
CUDA Archs: Not Set; ROCm: Disabled; Neuron: Disabled
GPU Topology:
GPU0 GPU1 CPU Affinity NUMA Affinity GPU NUMA ID
GPU0 X PHB 0-251 0 N/A
GPU1 PHB X 0-251 0 N/A

Legend:

X = Self
SYS = Connection traversing PCIe as well as the SMP interconnect between NUMA nodes (e.g., QPI/UPI)
NODE = Connection traversing PCIe as well as the interconnect between PCIe Host Bridges within a NUMA node
PHB = Connection traversing PCIe as well as a PCIe Host Bridge (typically the CPU)
PXB = Connection traversing multiple PCIe bridges (without traversing the PCIe Host Bridge)
PIX = Connection traversing at most a single PCIe bridge
NV# = Connection traversing a bonded set of # NVLinks

🐛 Describe the bug

I'm trying to run llama 3 70 b instruct with lora,

this is my start command:

python -m vllm.entrypoints.openai.api_server --model meta-llama/Meta-Llama-3-70B-Instruct --enable-lora --lora-modules stealthlora=./lora --max-lora-rank 64 --tensor-parallel-size 2

But I'm getting this error on inference:

File "/usr/local/lib/python3.10/dist-packages/vllm/worker/worker_base.py", line 146, in execute_method
raise e
File "/usr/local/lib/python3.10/dist-packages/vllm/worker/worker_base.py", line 137, in execute_method
return executor(*args, **kwargs)
File "/usr/local/lib/python3.10/dist-packages/torch/utils/_contextlib.py", line 115, in decorate_context
return func(*args, **kwargs)
File "/usr/local/lib/python3.10/dist-packages/vllm/worker/worker.py", line 249, in execute_model
output = self.model_runner.execute_model(seq_group_metadata_list,
File "/usr/local/lib/python3.10/dist-packages/torch/utils/_contextlib.py", line 115, in decorate_context
return func(*args, **kwargs)
File "/usr/local/lib/python3.10/dist-packages/vllm/worker/model_runner.py", line 790, in execute_model
self.set_active_loras(lora_requests, lora_mapping)
File "/usr/local/lib/python3.10/dist-packages/vllm/worker/model_runner.py", line 901, in set_active_loras
self.lora_manager.set_active_loras(lora_requests, lora_mapping)
File "/usr/local/lib/python3.10/dist-packages/vllm/lora/worker_manager.py", line 113, in set_active_loras
self._apply_loras(lora_requests)
File "/usr/local/lib/python3.10/dist-packages/vllm/lora/worker_manager.py", line 235, in _apply_loras
self.add_lora(lora)
File "/usr/local/lib/python3.10/dist-packages/vllm/lora/worker_manager.py", line 250, in add_lora
self._lora_manager.activate_lora(lora_request.lora_int_id)
File "/usr/local/lib/python3.10/dist-packages/vllm/lora/models.py", line 615, in activate_lora
result = super().activate_lora(lora_id)
File "/usr/local/lib/python3.10/dist-packages/vllm/lora/models.py", line 355, in activate_lora
module.set_lora(index, module_lora.lora_a, module_lora.lora_b,
File "/usr/local/lib/python3.10/dist-packages/vllm/lora/layers.py", line 800, in set_lora
lora_b = self.slice_lora_b(lora_b)
File "/usr/local/lib/python3.10/dist-packages/vllm/lora/layers.py", line 786, in slice_lora_b
lora_b = [lora_b_q, lora_b_k, lora_b_v]
UnboundLocalError: local variable 'lora_b_k' referenced before assignment

What could be the issue? I tried with mistral 7b lora, its working fine, but with llama3 70 b I get this error.

[DarkLight1337]

Should be fixed in the next release, by #4609.

[Stealthwriter]

Thanks, do you know when? or do you have an idea how I can go around this? It's happening when I run lora on a 70b model which is running on 2 GPU, I'm trying to load llama3 70b on a single GPU (a100) but it doesn't seem to work, I run our of vram...

[DarkLight1337]

The next release is just around the corner. If you can't wait for that, you can install vLLM from main branch directly.

[DarkLight1337]

Fixed by #4609, which has been released in v0.4.3.

Edit: Technically it is still in pre-release but should be out very soon.