A high-throughput and memory-efficient inference and serving engine for LLMs
Mirrored from https://github.com/tduncan-lum/Nvidia-Optimization-Stack---VLLM.git
Synced Nov 15, 11:00 UTC
[Bugfix] Fix LoRA test (#18123)
Signed-off-by: Jee Jee Li <pandaleefree@gmail.com>
Jee Jee Li committed
May 14, 2025 at 18:25 UTC
259127f8b8a8ee3c6b2ca7d91f2406f50b480f5f
1 file changed
+11
-5
tests/lora/test_worker.py
+11
-5
@@ -58,13 +58,19 @@ def test_worker_apply_lora(sql_lora_files):
58
download_dir=None,
59
load_format="dummy",
60
),
61
- parallel_config=ParallelConfig(1, 1, False),
61
+ parallel_config=ParallelConfig(
62
+ pipeline_parallel_size=1,
63
+ tensor_parallel_size=1,
64
+ data_parallel_size=1,
65
+ ),
66
scheduler_config=SchedulerConfig("generate", 32, 32, 32),
67
device_config=DeviceConfig("cuda"),
64
- cache_config=CacheConfig(block_size=16,
65
- gpu_memory_utilization=1.,
66
- swap_space=0,
67
- cache_dtype="auto"),
68
+ cache_config=CacheConfig(
69
+ block_size=16,
70
+ gpu_memory_utilization=1.0,
71
+ swap_space=0,
72
+ cache_dtype="auto",
73
+ ),
74
lora_config=LoRAConfig(max_lora_rank=8, max_cpu_loras=32,
75
max_loras=32),
76
)