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 17, 04:15 UTC
hotfix isort on logprobs ranks pr (#3622)
Simon Mo committed
Mar 25, 2024 at 11:55 UTC
f408d05c523c25e2f638a13cb34a2dab3dcb2754
2 files changed
+2
-1
tests/samplers/test_ranks.py
+1
@@ -1,4 +1,5 @@
1
import pytest
2
+
3
from vllm import SamplingParams
4
5
MODELS = ["facebook/opt-125m"]
tests/samplers/test_sampler.py
+1
-1
@@ -9,8 +9,8 @@ from transformers import GenerationConfig, GenerationMixin
9
from vllm.model_executor.layers.sampler import Sampler
10
from vllm.model_executor.utils import set_random_seed
11
from vllm.sequence import SamplingParams, SequenceData, SequenceGroupMetadata
12
-from vllm.worker.model_runner import ModelRunner
12
from vllm.utils import Counter
13
+from vllm.worker.model_runner import ModelRunner
14
15
16
class MockLogitsSampler(Sampler):