Merge tag 'v0.8.2' into v0.8.2-dev

pyproject.toml

 [build-system]
-# Should be mirrored in requirements-build.txt
+# Should be mirrored in requirements/build.txt
 requires = [
     "cmake>=3.26",
     "ninja",
     "packaging",
     "setuptools>=61",
     "setuptools-scm>=8.0",
-    "torch == 2.5.1",
+    "torch == 2.6.0",
     "wheel",
     "jinja2",
 ]
@@ -63,8 +63,30 @@ exclude = [
 ]
 
 [tool.ruff.lint.per-file-ignores]
+"vllm/third_party/**" = ["ALL"]
 "vllm/version.py" = ["F401"]
 "vllm/_version.py" = ["ALL"]
+# Python 3.8 typing. TODO: Remove these excludes after v1.0.0
+"vllm/adapter_commons/**/*.py" = ["UP006", "UP035"]
+"vllm/attention/**/*.py" = ["UP006", "UP035"]
+"vllm/compilation/**/*.py" = ["UP006", "UP035"]
+"vllm/core/**/*.py" = ["UP006", "UP035"]
+"vllm/device_allocator/**/*.py" = ["UP006", "UP035"]
+"vllm/distributed/**/*.py" = ["UP006", "UP035"]
+"vllm/engine/**/*.py" = ["UP006", "UP035"]
+"vllm/executor/**/*.py" = ["UP006", "UP035"]
+"vllm/lora/**/*.py" = ["UP006", "UP035"]
+"vllm/model_executor/**/*.py" = ["UP006", "UP035"]
+"vllm/platforms/**/*.py" = ["UP006", "UP035"]
+"vllm/plugins/**/*.py" = ["UP006", "UP035"]
+"vllm/profiler/**/*.py" = ["UP006", "UP035"]
+"vllm/prompt_adapter/**/*.py" = ["UP006", "UP035"]
+"vllm/spec_decode/**/*.py" = ["UP006", "UP035"]
+"vllm/transformers_utils/**/*.py" = ["UP006", "UP035"]
+"vllm/triton_utils/**/*.py" = ["UP006", "UP035"]
+"vllm/vllm_flash_attn/**/*.py" = ["UP006", "UP035"]
+"vllm/worker/**/*.py" = ["UP006", "UP035"]
+"vllm/utils.py" = ["UP006", "UP035"]
 
 [tool.ruff.lint]
 select = [
@@ -91,8 +113,6 @@ ignore = [
     "B007",
     # f-string format
     "UP032",
-    # Python 3.8 typing
-    "UP006", "UP035",
     # Can remove once 3.10+ is the minimum Python version
     "UP007",
 ]

requirements-cuda.txt

-# Common dependencies
--r requirements-common.txt
-
-# Dependencies for NVIDIA GPUs
-ray[adag] == 2.40.0 # Required for pipeline parallelism in V1.
-torch == 2.5.1
-torchaudio==2.5.1
-# These must be updated alongside torch
-torchvision == 0.20.1 # Required for phi3v processor. See https://github.com/pytorch/vision?tab=readme-ov-file#installation for corresponding version
-xformers == 0.0.28.post3; platform_system == 'Linux' and platform_machine == 'x86_64'  # Requires PyTorch 2.5.1

requirements-openvino.txt

-# Common dependencies
--r requirements-common.txt
-
-torch == 2.5.1 #  should be aligned with "common" vLLM torch version
-openvino >= 2024.4.0 # since 2024.4.0 both CPU and GPU support Paged Attention
-
-optimum @ git+https://github.com/huggingface/optimum.git # latest optimum is used to support latest transformers version
-optimum-intel[nncf] @ git+https://github.com/huggingface/optimum-intel.git # latest optimum-intel is used to support latest transformers version

requirements-xpu.txt

-# Common dependencies
--r requirements-common.txt
-
-ray >= 2.9
-cmake>=3.26
-ninja
-packaging
-setuptools-scm>=8
-wheel
-jinja2
-
-torch @ https://intel-optimized-pytorch.s3.cn-north-1.amazonaws.com.cn/ipex_dev/xpu/torch-2.5.0a0%2Bgite84e33f-cp310-cp310-linux_x86_64.whl
-intel-extension-for-pytorch @ https://intel-optimized-pytorch.s3.cn-north-1.amazonaws.com.cn/ipex_dev/xpu/intel_extension_for_pytorch-2.5.10%2Bgit9d489a8-cp310-cp310-linux_x86_64.whl
-oneccl_bind_pt @ https://intel-optimized-pytorch.s3.cn-north-1.amazonaws.com.cn/ipex_dev/xpu/oneccl_bind_pt-2.5.0%2Bxpu-cp310-cp310-linux_x86_64.whl
-
-triton-xpu == 3.0.0b1

requirements-build.txt → requirements/build.txt

@@ -4,6 +4,6 @@ ninja
 packaging
 setuptools>=61
 setuptools-scm>=8
-torch==2.5.1
+torch==2.6.0
 wheel
-jinja2
+jinja2>=3.1.6

requirements-common.txt → requirements/common.txt

+cachetools
 psutil
 sentencepiece  # Required for LLaMA tokenizer.
 numpy < 2.0.0
-numba == 0.60.0 # v0.61 doesn't support Python 3.9. Required for N-gram speculative decoding.
 requests >= 2.26.0
 tqdm
 blake3
@@ -9,8 +9,7 @@ py-cpuinfo
 transformers >= 4.49.0  # Required for Bamba model and Transformers backend.
 tokenizers >= 0.19.1  # Required for Llama 3.
 protobuf # Required by LlamaTokenizer.
-fastapi[standard] >= 0.107.0, < 0.113.0; python_version < '3.9'
-fastapi[standard]  >= 0.107.0, != 0.113.*, != 0.114.0; python_version >= '3.9'
+fastapi[standard] >= 0.115.0 # Required by FastAPI's form models in the OpenAI API server's audio transcriptions endpoint.
 aiohttp
 openai >= 1.52.0 # Ensure modern openai package (ensure types module present and max_completion_tokens field support)
 pydantic >= 2.9
@@ -18,10 +17,11 @@ prometheus_client >= 0.18.0
 pillow  # Required for image processing
 prometheus-fastapi-instrumentator >= 7.0.0
 tiktoken >= 0.6.0  # Required for DBRX tokenizer
-lm-format-enforcer >= 0.10.9, < 0.11
+lm-format-enforcer >= 0.10.11, < 0.11
+llguidance >= 0.7.9, < 0.8.0; platform_machine == "x86_64" or platform_machine == "arm64" or platform_machine == "aarch64"
 outlines == 0.1.11
-lark == 1.2.2 
-xgrammar == 0.1.11; platform_machine == "x86_64"
+lark == 1.2.2
+xgrammar == 0.1.16; platform_machine == "x86_64" or platform_machine == "aarch64"
 typing_extensions >= 4.10
 filelock >= 3.16.1 # need to contain https://github.com/tox-dev/filelock/pull/317
 partial-json-parser # used for parsing partial JSON outputs
@@ -29,11 +29,15 @@ pyzmq
 msgspec
 gguf == 0.10.0
 importlib_metadata
-mistral_common[opencv] >= 1.5.0
+mistral_common[opencv] >= 1.5.4
 pyyaml
 six>=1.16.0; python_version > '3.11' # transitive dependency of pandas that needs to be the latest version for python 3.12
 setuptools>=74.1.1; python_version > '3.11' # Setuptools is used by triton, we need to ensure a modern version is installed for 3.12+ so that it does not try to import distutils, which was removed in 3.12
 einops # Required for Qwen2-VL.
-compressed-tensors == 0.9.1 # required for compressed-tensors
+compressed-tensors == 0.9.2 # required for compressed-tensors
 depyf==0.18.0 # required for profiling and debugging with compilation config
 cloudpickle # allows pickling lambda functions in model_executor/models/registry.py
+watchfiles # required for http server to monitor the updates of TLS files
+python-json-logger # Used by logging as per examples/other/logging_configuration.md
+scipy # Required for phi-4-multimodal-instruct
+ninja # Required for xgrammar, rocm, tpu, xpu

requirements-cpu.txt → requirements/cpu.txt

 # Common dependencies
--r requirements-common.txt
+-r common.txt
 
 # Dependencies for CPUs
-torch==2.5.1+cpu; platform_machine != "ppc64le" and platform_machine != "aarch64" and platform_system != "Darwin"
-torch==2.5.1; platform_machine == "ppc64le" or platform_machine == "aarch64" or platform_system == "Darwin" 
+torch==2.6.0+cpu; platform_machine == "x86_64"
+torch==2.6.0; platform_system == "Darwin"
+torch==2.5.1; platform_machine == "ppc64le" or platform_machine == "aarch64"
+torch==2.7.0.dev20250304; platform_machine == "s390x"
 
 # required for the image processor of minicpm-o-2_6, this must be updated alongside torch
-torchaudio; platform_machine != "ppc64le"
+torchaudio; platform_machine != "ppc64le" and platform_machine != "s390x"
 torchaudio==2.5.1; platform_machine == "ppc64le"
 
 # required for the image processor of phi3v, this must be updated alongside torch
-torchvision; platform_machine != "ppc64le"
+torchvision; platform_machine != "ppc64le"  and platform_machine != "s390x"
 torchvision==0.20.1; platform_machine == "ppc64le"
 datasets # for benchmark scripts

requirements/cuda.txt

+# Common dependencies
+-r common.txt
+
+numba == 0.60.0 # v0.61 doesn't support Python 3.9. Required for N-gram speculative decoding
+
+# Dependencies for NVIDIA GPUs
+ray[cgraph]>=2.43.0 # Ray Compiled Graph, required for pipeline parallelism in V1.
+torch==2.6.0
+torchaudio==2.6.0
+# These must be updated alongside torch
+torchvision==0.21.0 # Required for phi3v processor. See https://github.com/pytorch/vision?tab=readme-ov-file#installation for corresponding version
+xformers==0.0.29.post2; platform_system == 'Linux' and platform_machine == 'x86_64'  # Requires PyTorch 2.6.0

requirements-dev.txt → requirements/dev.txt

--r requirements-lint.txt
--r requirements-test.txt
+-r lint.txt
+-r test.txt
 
 # Avoid adding requirements directly to this file.
 # Instead, modify the two files referenced above.

docs/requirements-docs.txt → requirements/docs.txt

@@ -9,12 +9,13 @@ msgspec
 cloudpickle
 
 # packages to install to build the documentation
+cachetools
 pydantic >= 2.8
 -f https://download.pytorch.org/whl/cpu
 torch
 py-cpuinfo
 transformers
-mistral_common >= 1.5.0
+mistral_common >= 1.5.4
 aiohttp
 starlette
 openai # Required by docs/source/serving/openai_compatible_server.md's vllm.entrypoints.openai.cli_args

requirements-hpu.txt → requirements/hpu.txt

 # Common dependencies
--r requirements-common.txt
+-r common.txt
 
 # Dependencies for HPU code
 ray

requirements-neuron.txt → requirements/neuron.txt

 # Common dependencies
--r requirements-common.txt
+-r common.txt
 
 # Dependencies for Neuron devices
 torch-neuronx >= 2.5.0

requirements-rocm-build.txt → requirements/rocm-build.txt

 # Common dependencies
--r requirements-common.txt
+-r common.txt
 
---extra-index-url https://download.pytorch.org/whl/rocm6.2
-torch==2.5.1
-torchvision==0.20.1
-torchaudio==2.5.1
+--extra-index-url https://download.pytorch.org/whl/rocm6.2.4
+torch==2.6.0
+torchvision==0.21.0
+torchaudio==2.6.0
 
 cmake>=3.26
-ninja
 packaging
 setuptools>=61
 setuptools-scm>=8
 wheel
-jinja2
+jinja2>=3.1.6
 amdsmi==6.2.4

requirements/rocm-test.txt

+
+# entrypoints test
+# librosa==0.10.2.post1 # required by audio tests in entrypoints/openai
+audioread==3.0.1
+cffi==1.17.1
+decorator==5.2.1
+lazy-loader==0.4
+platformdirs==4.3.6
+pooch==1.8.2
+#pycparse==2.22
+soundfile==0.13.1
+soxr==0.5.0.post1
+librosa==0.10.2.post1
+
+# entrypoints test
+#vllm[video] # required by entrypoints/openai/test_video.py
+decord==0.6.0
+
+# entrypoints test
+#sentence-transformers # required by entrypoints/openai/test_score.py
+sentence-transformers==3.4.1
+
+

requirements-rocm.txt → requirements/rocm.txt

 # Common dependencies
--r requirements-common.txt
+-r common.txt
+
+numba == 0.60.0 # v0.61 doesn't support Python 3.9. Required for N-gram speculative decoding
 
 # Dependencies for hcus
 awscli

requirements-test.in → requirements/test.in

@@ -8,6 +8,7 @@ pytest-shard
 
 # testing utils
 awscli
+backoff # required for phi4mm test
 decord # required for video tests
 einops # required for MPT, qwen-vl and Mamba
 httpx
@@ -16,21 +17,22 @@ vector_quantize_pytorch # required for minicpmo_26 test
 vocos # required for minicpmo_26 test
 peft
 pqdm
-ray[adag]==2.40.0
+ray[cgraph]>=2.43.0 # Ray Compiled Graph, required by pipeline parallelism tests
 sentence-transformers # required for embedding tests
 soundfile # required for audio tests
 jiwer # required for audio tests
 timm # required for internvl test
-torch==2.5.1
-torchaudio==2.5.1
+torch==2.6.0
+torchaudio==2.6.0
+torchvision==0.21.0
 transformers_stream_generator # required for qwen-vl test
 matplotlib # required for qwen-vl test
-mistral_common[opencv] >= 1.5.0 # required for pixtral test
+mistral_common[opencv] >= 1.5.4 # required for pixtral test
 datamodel_code_generator # required for minicpm3 test
 lm-eval[api]==0.4.4 # required for model evaluation test
-transformers==4.48.2 
+transformers==4.48.2
 # quantization
-bitsandbytes>=0.45.0
+bitsandbytes>=0.45.3
 buildkite-test-collector==0.1.9
 
 genai_perf==0.0.8
@@ -38,4 +40,5 @@ tritonclient==2.51.0
 
 numpy < 2.0.0
 runai-model-streamer==0.11.0
-runai-model-streamer-s3==0.11.0
+runai-model-streamer-s3==0.11.0
\ No newline at end of file
+fastsafetensors>=0.1.10

requirements-test.txt → requirements/test.txt

-#
-# This file is autogenerated by pip-compile with Python 3.12
-# by the following command:
-#
-# python3.12 -m piptools compile requirements-test.in -o requirements-test.txt
-#
+# This file was autogenerated by uv via the following command:
+#    uv pip compile requirements/test.in -o requirements/test.txt
 absl-py==2.1.0
     # via rouge-score
 accelerate==1.0.1
@@ -12,7 +8,7 @@ accelerate==1.0.1
     #   peft
 aiohappyeyeballs==2.4.3
     # via aiohttp
-aiohttp==3.10.10
+aiohttp==3.10.11
     # via
     #   datasets
     #   fsspec
@@ -36,9 +32,11 @@ attrs==24.2.0
 audioread==3.0.1
     # via librosa
 awscli==1.35.23
-    # via -r requirements-test.in
-bitsandbytes==0.45.0
-    # via -r requirements-test.in
+    # via -r requirements/test.in
+backoff==2.2.1
+    # via -r requirements/test.in
+bitsandbytes==0.45.3
+    # via -r requirements/test.in
 black==24.10.0
     # via datamodel-code-generator
 boto3==1.35.57
@@ -51,7 +49,7 @@ botocore==1.35.57
 bounded-pool-executor==0.0.3
     # via pqdm
 buildkite-test-collector==0.1.9
-    # via -r requirements-test.in
+    # via -r requirements/test.in
 certifi==2024.8.30
     # via
     #   httpcore
@@ -69,6 +67,7 @@ click==8.1.7
     #   jiwer
     #   nltk
     #   ray
+    #   typer
 colorama==0.4.6
     # via
     #   awscli
@@ -83,7 +82,7 @@ cupy-cuda12x==13.3.0
 cycler==0.12.1
     # via matplotlib
 datamodel-code-generator==0.26.3
-    # via -r requirements-test.in
+    # via -r requirements/test.in
 dataproperty==1.0.1
     # via
     #   pytablewriter
@@ -95,7 +94,7 @@ datasets==3.0.2
 decorator==5.1.1
     # via librosa
 decord==0.6.0
-    # via -r requirements-test.in
+    # via -r requirements/test.in
 dill==0.3.8
     # via
     #   datasets
@@ -108,7 +107,7 @@ docutils==0.16
     # via awscli
 einops==0.8.0
     # via
-    #   -r requirements-test.in
+    #   -r requirements/test.in
     #   encodec
     #   vector-quantize-pytorch
     #   vocos
@@ -124,6 +123,8 @@ fastparquet==2024.11.0
     # via genai-perf
 fastrlock==0.8.2
     # via cupy-cuda12x
+fastsafetensors==0.1.10
+    # via -r requirements/test.in
 filelock==3.16.1
     # via
     #   datasets
@@ -131,7 +132,6 @@ filelock==3.16.1
     #   ray
     #   torch
     #   transformers
-    #   triton
 fonttools==4.54.1
     # via matplotlib
 frozendict==2.4.6
@@ -141,7 +141,7 @@ frozenlist==1.5.0
     #   aiohttp
     #   aiosignal
     #   ray
-fsspec[http]==2024.9.0
+fsspec==2024.9.0
     # via
     #   datasets
     #   evaluate
@@ -149,7 +149,7 @@ fsspec[http]==2024.9.0
     #   huggingface-hub
     #   torch
 genai-perf==0.0.8
-    # via -r requirements-test.in
+    # via -r requirements/test.in
 genson==1.3.0
     # via datamodel-code-generator
 h11==0.14.0
@@ -159,7 +159,7 @@ hiredis==3.0.0
 httpcore==1.0.6
     # via httpx
 httpx==0.27.2
-    # via -r requirements-test.in
+    # via -r requirements/test.in
 huggingface-hub==0.26.2
     # via
     #   accelerate
@@ -186,12 +186,12 @@ iniconfig==2.0.0
     # via pytest
 isort==5.13.2
     # via datamodel-code-generator
-jinja2==3.1.4
+jinja2==3.1.6
     # via
     #   datamodel-code-generator
     #   torch
 jiwer==3.0.5
-    # via -r requirements-test.in
+    # via -r requirements/test.in
 jmespath==1.0.1
     # via
     #   boto3
@@ -218,11 +218,11 @@ lazy-loader==0.4
 libnacl==2.1.0
     # via tensorizer
 librosa==0.10.2.post1
-    # via -r requirements-test.in
+    # via -r requirements/test.in
 llvmlite==0.43.0
     # via numba
-lm-eval[api]==0.4.4
-    # via -r requirements-test.in
+lm-eval==0.4.4
+    # via -r requirements/test.in
 lxml==5.3.0
     # via sacrebleu
 markdown-it-py==3.0.0
@@ -230,7 +230,7 @@ markdown-it-py==3.0.0
 markupsafe==3.0.2
     # via jinja2
 matplotlib==3.9.2
-    # via -r requirements-test.in
+    # via -r requirements/test.in
 mbstrdecoder==1.1.3
     # via
     #   dataproperty
@@ -238,10 +238,8 @@ mbstrdecoder==1.1.3
     #   typepy
 mdurl==0.1.2
     # via markdown-it-py
-mistral-common[opencv]==1.5.1
-    # via
-    #   -r requirements-test.in
-    #   mistral-common
+mistral-common==1.5.4
+    # via -r requirements/test.in
 more-itertools==10.5.0
     # via lm-eval
 mpmath==1.3.0
@@ -270,7 +268,7 @@ numexpr==2.10.1
     # via lm-eval
 numpy==1.26.4
     # via
-    #   -r requirements-test.in
+    #   -r requirements/test.in
     #   accelerate
     #   bitsandbytes
     #   contourpy
@@ -326,6 +324,8 @@ nvidia-cusparse-cu12==12.3.1.170
     # via
     #   nvidia-cusolver-cu12
     #   torch
+nvidia-cusparselt-cu12==0.6.2
+    # via torch
 nvidia-nccl-cu12==2.21.5
     # via torch
 nvidia-nvjitlink-cu12==12.4.127
@@ -372,7 +372,7 @@ patsy==1.0.1
     # via statsmodels
 peft==0.13.2
     # via
-    #   -r requirements-test.in
+    #   -r requirements/test.in
     #   lm-eval
 pillow==10.4.0
     # via
@@ -394,7 +394,7 @@ pooch==1.8.2
 portalocker==2.10.1
     # via sacrebleu
 pqdm==0.2.0
-    # via -r requirements-test.in
+    # via -r requirements/test.in
 propcache==0.2.0
     # via yarl
 protobuf==5.28.3
@@ -418,7 +418,7 @@ pybind11==2.13.6
     # via lm-eval
 pycparser==2.22
     # via cffi
-pydantic[email]==2.9.2
+pydantic==2.9.2
     # via
     #   datamodel-code-generator
     #   mistral-common
@@ -432,7 +432,7 @@ pytablewriter==1.2.0
     # via lm-eval
 pytest==8.3.3
     # via
-    #   -r requirements-test.in
+    #   -r requirements/test.in
     #   buildkite-test-collector
     #   genai-perf
     #   pytest-asyncio
@@ -441,15 +441,15 @@ pytest==8.3.3
     #   pytest-rerunfailures
     #   pytest-shard
 pytest-asyncio==0.24.0
-    # via -r requirements-test.in
+    # via -r requirements/test.in
 pytest-forked==1.6.0
-    # via -r requirements-test.in
+    # via -r requirements/test.in
 pytest-mock==3.14.0
     # via genai-perf
 pytest-rerunfailures==14.0
-    # via -r requirements-test.in
+    # via -r requirements/test.in
 pytest-shard==0.1.2
-    # via -r requirements-test.in
+    # via -r requirements/test.in
 python-dateutil==2.9.0.post0
     # via
     #   botocore
@@ -478,8 +478,8 @@ pyyaml==6.0.2
     #   vocos
 rapidfuzz==3.12.1
     # via jiwer
-ray[adag]==2.40.0
-    # via -r requirements-test.in
+ray==2.43.0
+    # via -r requirements/test.in
 redis==5.2.0
     # via tensorizer
 referencing==0.35.1
@@ -508,7 +508,9 @@ requests==2.32.3
 responses==0.25.3
     # via genai-perf
 rich==13.9.4
-    # via genai-perf
+    # via
+    #   genai-perf
+    #   typer
 rouge-score==0.1.2
     # via lm-eval
 rpds-py==0.20.1
@@ -518,9 +520,9 @@ rpds-py==0.20.1
 rsa==4.7.2
     # via awscli
 runai-model-streamer==0.11.0
-    # via -r requirements-test.in
+    # via -r requirements/test.in
 runai-model-streamer-s3==0.11.0
-    # via -r requirements-test.in
+    # via -r requirements/test.in
 s3transfer==0.10.3
     # via
     #   awscli
@@ -546,9 +548,15 @@ scipy==1.13.1
     #   statsmodels
     #   vocos
 sentence-transformers==3.2.1
-    # via -r requirements-test.in
+    # via -r requirements/test.in
 sentencepiece==0.2.0
     # via mistral-common
+setuptools==75.8.0
+    # via
+    #   pytablewriter
+    #   torch
+shellingham==1.5.4
+    # via typer
 six==1.16.0
     # via
     #   python-dateutil
@@ -559,7 +567,7 @@ sniffio==1.3.1
     #   httpx
 soundfile==0.12.1
     # via
-    #   -r requirements-test.in
+    #   -r requirements/test.in
     #   librosa
 soxr==0.5.0.post1
     # via librosa
@@ -582,7 +590,7 @@ tenacity==9.0.0
     #   lm-eval
     #   plotly
 tensorizer==2.9.0
-    # via -r requirements-test.in
+    # via -r requirements/test.in
 threadpoolctl==3.5.0
     # via scikit-learn
 tiktoken==0.7.0
@@ -590,15 +598,16 @@ tiktoken==0.7.0
     #   lm-eval
     #   mistral-common
 timm==1.0.11
-    # via -r requirements-test.in
+    # via -r requirements/test.in
 tokenizers==0.21.0
     # via transformers
-torch==2.5.1
+torch==2.6.0
     # via
-    #   -r requirements-test.in
+    #   -r requirements/test.in
     #   accelerate
     #   bitsandbytes
     #   encodec
+    #   fastsafetensors
     #   lm-eval
     #   peft
     #   runai-model-streamer
@@ -609,13 +618,15 @@ torch==2.5.1
     #   torchvision
     #   vector-quantize-pytorch
     #   vocos
-torchaudio==2.5.1
+torchaudio==2.6.0
     # via
-    #   -r requirements-test.in
+    #   -r requirements/test.in
     #   encodec
     #   vocos
-torchvision==0.20.1
-    # via timm
+torchvision==0.21.0
+    # via
+    #   -r requirements/test.in
+    #   timm
 tqdm==4.66.6
     # via
     #   datasets
@@ -632,28 +643,29 @@ tqdm-multiprocess==0.0.11
     # via lm-eval
 transformers==4.48.2
     # via
-    #   -r requirements-test.in
+    #   -r requirements/test.in
     #   genai-perf
     #   lm-eval
     #   peft
     #   sentence-transformers
     #   transformers-stream-generator
 transformers-stream-generator==0.0.5
-    # via -r requirements-test.in
-triton==3.1.0
+    # via -r requirements/test.in
+triton==3.2.0
     # via torch
 tritonclient==2.51.0
     # via
-    #   -r requirements-test.in
+    #   -r requirements/test.in
     #   genai-perf
-typepy[datetime]==1.3.2
+typepy==1.3.2
     # via
     #   dataproperty
     #   pytablewriter
     #   tabledata
+typer==0.15.2
+    # via fastsafetensors
 typing-extensions==4.12.2
     # via
-    #   bitsandbytes
     #   huggingface-hub
     #   librosa
     #   mistral-common
@@ -661,6 +673,7 @@ typing-extensions==4.12.2
     #   pydantic
     #   pydantic-core
     #   torch
+    #   typer
 tzdata==2024.2
     # via pandas
 urllib3==2.2.3
@@ -670,9 +683,9 @@ urllib3==2.2.3
     #   responses
     #   tritonclient
 vector-quantize-pytorch==1.21.2
-    # via -r requirements-test.in
+    # via -r requirements/test.in
 vocos==0.1.0
-    # via -r requirements-test.in
+    # via -r requirements/test.in
 word2number==1.1
     # via lm-eval
 xxhash==3.5.0
@@ -683,6 +696,3 @@ yarl==1.17.1
     # via aiohttp
 zstandard==0.23.0
     # via lm-eval
-
-# The following packages are considered to be unsafe in a requirements file:
-# setuptools
\ No newline at end of file

requirements-tpu.txt → requirements/tpu.txt

 # Common dependencies
--r requirements-common.txt
+-r common.txt
 
 # Dependencies for TPU
 cmake>=3.26
-ninja
 packaging
 setuptools-scm>=8
 wheel
-jinja2
+jinja2>=3.1.6
 ray[default]
+ray[data]
 
 # Install torch_xla
 --pre
@@ -17,7 +17,9 @@ ray[default]
 --find-links https://storage.googleapis.com/libtpu-releases/index.html
 --find-links https://storage.googleapis.com/jax-releases/jax_nightly_releases.html
 --find-links https://storage.googleapis.com/jax-releases/jaxlib_nightly_releases.html
-torch==2.6.0.dev20241216+cpu
-torch_xla[tpu, pallas] @ https://storage.googleapis.com/pytorch-xla-releases/wheels/tpuvm/torch_xla-2.7.0.dev20250124-cp39-cp39-linux_x86_64.whl ; python_version == "3.9"
-torch_xla[tpu, pallas] @ https://storage.googleapis.com/pytorch-xla-releases/wheels/tpuvm/torch_xla-2.7.0.dev20250124-cp310-cp310-linux_x86_64.whl ; python_version == "3.10"
-torch_xla[tpu, pallas] @ https://storage.googleapis.com/pytorch-xla-releases/wheels/tpuvm/torch_xla-2.7.0.dev20250124-cp311-cp311-linux_x86_64.whl ; python_version == "3.11"
+torch @ https://storage.googleapis.com/pytorch-xla-releases/wheels/tpuvm/torch-2.8.0.dev20250319-cp39-cp39-linux_x86_64.whl ; python_version == "3.9"
+torch @ https://storage.googleapis.com/pytorch-xla-releases/wheels/tpuvm/torch-2.8.0.dev20250319-cp310-cp310-linux_x86_64.whl ; python_version == "3.10"
+torch @ https://storage.googleapis.com/pytorch-xla-releases/wheels/tpuvm/torch-2.8.0.dev20250319-cp311-cp311-linux_x86_64.whl ; python_version == "3.11"
+torch_xla[tpu, pallas] @ https://storage.googleapis.com/pytorch-xla-releases/wheels/tpuvm/torch_xla-2.8.0.dev20250319-cp39-cp39-linux_x86_64.whl ; python_version == "3.9"
+torch_xla[tpu, pallas] @ https://storage.googleapis.com/pytorch-xla-releases/wheels/tpuvm/torch_xla-2.8.0.dev20250319-cp310-cp310-linux_x86_64.whl ; python_version == "3.10"
+torch_xla[tpu, pallas] @ https://storage.googleapis.com/pytorch-xla-releases/wheels/tpuvm/torch_xla-2.8.0.dev20250319-cp311-cp311-linux_x86_64.whl ; python_version == "3.11"

requirements/xpu.txt

+# Common dependencies
+-r common.txt
+
+ray>=2.9
+cmake>=3.26
+packaging
+setuptools-scm>=8
+setuptools>=75.8.0
+wheel
+jinja2>=3.1.6
+datasets # for benchmark scripts
+
+torch==2.6.0+xpu
+torchaudio
+torchvision
+pytorch-triton-xpu
+--extra-index-url=https://download.pytorch.org/whl/xpu
+
+# Please refer xpu doc, we need manually install intel-extension-for-pytorch 2.6.10+xpu due to there are some conflict dependencies with torch 2.6.0+xpu
+# FIXME: This will be fix in ipex 2.7. just leave this here for awareness.
+# intel-extension-for-pytorch==2.6.10+xpu
+oneccl_bind_pt==2.6.0+xpu
+--extra-index-url=https://pytorch-extension.intel.com/release-whl/stable/xpu/us/