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
Add logging for torch nightly version (#17669)
Signed-off-by: Yang Wang <elainewy@meta.com>
Yang Wang committed
May 6, 2025 at 17:45 UTC
6de3e13413a526cf7b908455e8cc35665cb19ef0
2 files changed
+12
-2
docker/Dockerfile.nightly_torch
+3
-1
@@ -309,5 +309,7 @@ ENV HF_HUB_ENABLE_HF_TRANSFER 1
309
RUN --mount=type=cache,target=/root/.cache/uv \
310
uv pip install --system -r requirements/nightly_torch_test.txt
311
312
-#################### UNITTEST IMAGE #############################
312
+# Logging to confirm the torch versions
313
+RUN pip freeze | grep -E 'torch|xformers|vllm|flashinfer'
314
315
+#################### UNITTEST IMAGE #############################
requirements/nightly_torch_test.txt
+9
-1
@@ -8,7 +8,6 @@ pytest-rerunfailures
8
pytest-shard
9
pytest-timeout
10
11
-
11
librosa # required by audio tests in entrypoints/openai
12
sentence-transformers
13
numba == 0.61.2; python_version > '3.9'
@@ -31,3 +30,12 @@ bitsandbytes>=0.45.3
30
# required for minicpmo_26 test
31
vector_quantize_pytorch
32
vocos
33
+
34
+# required for Basic Models Test
35
+blobfile # required for kimi-vl test
36
+matplotlib # required for qwen-vl test
37
+
38
+# required for Multi-Modal Models Test (Standard)
39
+num2words # required for smolvlm test
40
+pqdm
41
+timm # required for internvl test