Ambiente de GPU serverless versão 6 (Prévia)

Importante

Esse recurso está em Visualização Pública.

Esta página descreve as informações do ambiente do sistema para o ambiente de GPU Serverless versão 6. Essa oferta de computação faz parte do AI Runtime, que foi projetado para cargas de trabalho modernas de IA e aprendizado profundo.

O ambiente 6 de GPU serverless é construído sobre o ambiente serverless 6 (CPU). Veja as novidades em ambiente sem servidor 6 (CPU). Ele inclui o seguinte ambiente:

Para garantir a compatibilidade com o aplicativo, as cargas de trabalho de GPU sem servidor usam uma API com versão, conhecida como a versão do ambiente, que permanece compatível com versões de servidor mais recentes.

Você pode selecionar um ambiente base que inclua essa versão de ambiente usando o painel lateral Ambiente em seus notebooks sem servidor. Consulte Interativo (Notebooks).

Novos recursos e melhorias

Na versão 6, as bibliotecas CUDA são atualizadas para NVIDIA CUDA 13, e as dependências associadas são atualizadas para corresponder, incluindo MLflow e a API Python da GPU Serverless. As bibliotecas comuns de aprendizado de máquina e IA no ambiente de IA Databricks foram atualizadas para compatibilidade com a nova versão CUDA, juntamente com a adição de várias bibliotecas recentemente populares.

CUDA 13 e PyTorch 2.11

Ambiente de IA Databricks

A versão 6 atualiza o ambiente de IA do Databricks para o kit de ferramentas NVIDIA CUDA 13 e a pilha de runtime. PyTorch é atualizado para torch 2.11.0 (construído para CUDA 13.0, +cu130) com torchvision 0.26.0, e as bibliotecas NVIDIA incluídas (cuBLAS, cuDNN, NCCL, cuFFT, cuSPARSE e componentes de runtime relacionados) são atualizadas para suas versões CUDA 13.

Atualizaram bibliotecas de deep learning e LLM

Ambiente de IA Databricks

Bibliotecas principais de aprendizado de máquina são atualizadas no ambiente de IA Databricks, incluindo:

  • transformers 5.8.1
  • vllm 0.24.0
  • ray 2.56.0
  • langchain 1.3.11
  • sentence-transformers 5.5.1
  • openai 2.37.0
  • scikit-learn 1.7.2
  • scipy 1.16.3
  • trl 0.24.0

API Python para GPU sem servidor

A API 0.5.24 Python para GPU Serverless está incluída nesse ambiente. Para o histórico completo das alterações na API Python de GPU Serverless, consulte as notas de versão 5.

Pacotes adicionados ao ambiente de IA do Databricks

Os seguintes pacotes notáveis foram adicionados ao ambiente de IA do Databricks na versão 6:

  • langgraph 1.2.8 (com langgraph-checkpoint, langgraph-prebuilt e langgraph-sdk)
  • langchain-classic 1.0.8
  • lightning 2.5.5
  • polars 1.42.1
  • ultralytics 8.4.90
  • flash_attn_3 3.0.0 e flash-attn-4 4.0.0b18
  • flashinfer-cubin 0.6.12
  • liger_kernel 0.8.1
  • fastsafetensors 0.3.2
  • tilelang 0.1.9
  • xgboost-cu13 3.2.0 (Construção do CUDA 13, substituindo xgboost)
  • tabfm 1.0.1

Além disso, o toolkit CUDA 13 e as bibliotecas de runtime (por exemplo, cuda-toolkit 13.0.2, nvidia-cudnn-cu13, nvidia-nccl-cu13 e pacotes nvidia-* relacionados) foram adicionados como parte da atualização para o CUDA 13.

Pacotes removidos do ambiente de IA do Databricks

Os seguintes pacotes notáveis foram removidos do ambiente de IA Databricks na versão 6:

  • unsloth (e unsloth_zoo)
  • xformers
  • diffusers
  • bitsandbytes
  • torchao
  • litellm
  • langchain-openai
  • cupy-cuda12x
  • mosaicml-streaming
  • gguf
  • seaborn
  • xgboost (substituído por xgboost-cu13) e xgboost-ray

As bibliotecas de tempo de execução do CUDA 12 (os nvidia-*-cu12 pacotes) foram removidas como parte da atualização do CUDA 13. A pilha de desenvolvimento do JupyterLab e de notebooks (por exemplo, jupyterlab, notebook, nbconvert, nbclient, nbformat) e as ferramentas de desenvolvimento do Python LSP (python-lsp-server, python-lsp-jsonrpc, rope, yapf) também não fazem mais parte do ambiente de IA da Databricks.

Pacotes não incluídos no ambiente base

Os seguintes pacotes não estão incluídos no ambiente base, mas permanecem disponíveis no ambiente de IA do Databricks:

  • flash_attn
  • torch
  • torchvision
  • scikit-learn
  • scipy

Com base no feedback dos clientes solicitando controle sobre o versionamento do PyTorch, torch não está incluído no ambiente base da versão 6. Isso permite que você instale a versão que melhor se ajusta à sua carga de trabalho. Para usar uma versão testada pelo Databricks, execute %pip install torch==2.11.0 em uma célula do notebook ou instale torch==2.11.0 usando a seção Dependências do painel lateral ambiente .

Ambiente do sistema

  • Sistema Operacional: Ubuntu 24.04 LTS
  • Python: 3.12.3
  • Databricks Connect: 19.0.0
  • Kit de ferramentas NVIDIA CUDA: 13.0

Bibliotecas do Python instaladas

Ambiente padrão

Para reproduzir o ambiente de GPU serverless 6 no seu ambiente virtual Python local, baixe o arquivorequirements-env-gpu-6.txt e execute pip install -r requirements-env-gpu-6.txt. Esse comando instala todas as bibliotecas do ambiente serverless 6.

Library Versão Library Versão Library Versão
documento-anotado 0.0.4 annotated-types 0.7.0 anyio 4.10.0
arro3-core 0.8.0 asttokens 3.0.0 astunparse 1.6.3
azure-core 1.41.0 azure-identity 1.25.3 azure-mgmt-core 1.6.0
azure-mgmt-web 10.1.0 azure-storage-blob 12.29.0 Serviço de armazenamento de arquivos do Azure chamado "azure-storage-file-datalake" 12.24.0
black 25.9.0 blinker 1.7.0 boto3 1.40.46
botocore 1.40.46 cachetools 5.5.1 certifi... 2025.11.12
cffi 2.0.0 charset-normalizer 3.4.4 clique 8.2.1
cloudpickle 3.1.1 comm 0.2.3 contourpy 1.3.3
criptografia 46.0.3 cycler 0.11.0 agentes do databricks 1.11.0
databricks-connect 19.0.0 databricks-sdk 0.122.0 databricks.serverless_gpu 0.5.24
dataclasses-json 0.6.7 dbus-python 1.3.2 debugpy 1.8.16
decorador 5.2.1 deltalake 1.5.1 Deprecated 1.3.1
distlib 0.4.0 executing 2.2.1 fastapi 0.136.3
filelock 3.29.0 fonttools 4.60.1 fsspec 2025.9.0
future 1.0.0 gitdb 4.0.11 GitPython 3.1.45
google-api-core 2.30.3 google-auth 2.53.0 google-cloud-core 2.6.0
google-cloud-storage 3.10.1 google-crc32c 1.8.0 google-resumable-media 2.9.0
googleapis-common-protos 1.71.0 grpcio 1.76.0 grpcio-status 1.76.0
h11 0.16.0 httplib2 0.20.4 idna 3.11
importlib_metadata 8.7.0 iniconfig 2.1.0 ipyflow-core 0.0.227
ipykernel 6.31.0 ipython 9.7.0 ipython_pygments_lexers 1.1.1
ipywidgets 8.1.7 isodate 0.7.2 jedi 0.19.2
Jinja2 3.1.6 jmespath 1.0.1 joblib 1.5.2
jupyter_client 8.6.3 jupyter_core 5.8.1 jupyterlab_widgets 3.0.15
kiwisolver 1.4.8 launchpadlib 1.11.0 lazr.restfulclient 0.14.6
lazr.uri 1.0.6 markdown-it-py 2.2.0 MarkupSafe 3.0.2
malvavisco 3.26.2 matplotlib 3.10.6 matplotlib-inline 0.2.1
mccabe 0.7.0 mdurl 0.1.2 mlflow-skinny 3.12.0
mmh3 5.2.1 msal 1.36.0 msal-extensions 1.3.1
msrest 0.7.1 mypy-extensions 1.0.0 nest-asyncio 1.6.0
nodeenv 1.10.0 numpy 2.3.4 nvidia-ml-py 13.590.44
oauthlib 3.2.2 opencv_python_headless 5.0.0.93 opentelemetry-api 1.42.1
opentelemetry-proto 1.42.1 opentelemetry-sdk 1.42.1 opentelemetry-semantic-conventions 0,63b1
packaging 25.0 pandas 2.3.3 parambench-train-comms 0.0.0
parso 0.8.5 pathspec 0.12.1 pexpect 4.9.0
pillow 12.0.0 pip 25,3 platformdirs 4.5.0
pluggy 1.5.0 prompt_toolkit 3.0.52 proto-plus 1.28.0
protobuf 6.33.5 psutil 7.0.0 ptyprocess 0.7.0
pure_eval 0.2.3 py4j 0.10.9.9 pyarrow 21.0.0
pyasn1 0.6.1 pyasn1_modules 0.4.2 pyccolo 0.0.83
pycparser 2.23 pydantic 2.13.3 pydantic_core 2.46.3
pydot 4.0.0 pyflakes 3.2.0 Pygments 2.19.2
PyGObject 3.48.2 pyiceberg 0.11.1 PyJWT 2.10.1
pyparsing 3.2.5 pyright 1.1.409 pyroaring 1.1.0
pyspark 4.3.0.dev0+databricks.connect.19.0.0 pytest 8.4.2 python-dateutil 2.9.0.post0
python-dotenv 1.2.2 pytokens 0.2.0 pytz 2025.2
PyYAML 6.0.3 pyzmq 27.1.0 regex 2025.9.1
requests 2.32.5 requests-oauthlib 2.0.0 rich 14.2.0
s3transfer 0.14.0 setuptools 80.9.0 six 1.17.0
smmap 5.0.0 sniffio 1.3.0 sqlparse 0.5.5
ssh-import-id 5.11 stack-data 0.6.3 starlette 0.52.1
strictyaml 1.7.3 tenacity 9.1.2 threadpoolctl 3.5.0
tiktoken 0.13.0 tokenize_rt 6.2.0 tomli 2.0.1
tornado 6.5.1 tqdm 4.67.1 traitlets 5.14.3
typing-inspect 0.9.0 inspeção de digitação 0.4.2 typing_extensions 4.15.0
tzdata 2026.2 unattended-upgrades 0.1 urllib3 2.5.0
uvicorn 0.48.0 virtualenv 20.35.4 wadllib 1.3.6
wcwidth 0.2.13 wheel 0.45.1 whenever 0.7.3
widgetsnbextension 4.0.14 wrapt 1.17.0 zipp 3.23.0
zstandard 0.25.0

Ambiente de IA

O ambiente de IA inclui todos os pacotes do ambiente Standard, bem como os seguintes pacotes. Esse ambiente inclui pyTorch e bibliotecas abrangentes de machine learning:

  • torch 2.11.0
  • torchvision 0.26.0
  • flash_attn 2.8.3
  • catboost 1.2.10
  • langchain 1.3.11
  • langgraph 1.2.8
  • lightgbm 4.6.0
  • lightning 2.5.5
  • openai 2.37.0
  • optuna 4.5.0
  • polars 1.42.1
  • pytorch-lightning 2.5.4
  • ray 2.56.0
  • scikit-learn 1.7.2
  • scipy 1.16.3
  • sentence-transformers 5.5.1
  • transformers 5.8.1
  • trl 0.24.0
  • ultralytics 8.4.90
  • vllm 0.24.0
  • xgboost-cu13 3.2.0

Para reproduzir o ambiente de IA V6 no seu ambiente virtual Python local, baixe o arquivorequirements-ai-gpu-6.txt e execute pip install -r requirements-ai-gpu-6.txt. Esse comando instala todas as bibliotecas do ambiente de IA.

Library Versão Library Versão Library Versão
absl-py 2.4.0 acelerar 1.13.0 aiohappyeyeballs 2.6.1
aiohttp 3.13.5 aiohttp-cors 0.8.1 aiosignal 1.4.0
alambique 1.18.5 documento-anotado 0.0.4 annotated-types 0.7.0
Antrópico 0.116.0 antlr4-python3-runtime 4.9.3 anyio 4.10.0
apache-tvm-ffi 0.1.9 arro3-core 0.8.0 astor 0.8.1
asttokens 3.0.0 astunparse 1.6.3 attrs 25.4.0
audioread 3.1.0 azure-core 1.41.0 azure-identity 1.25.3
azure-mgmt-core 1.6.0 azure-mgmt-web 10.1.0 azure-storage-blob 12.29.0
Serviço de armazenamento de arquivos do Azure chamado "azure-storage-file-datalake" 12.24.0 black 25.9.0 blake3 1.0.9
blinker 1.7.0 boto3 1.40.46 botocore 1.40.46
cachetools 5.5.1 catboost 1.2.10 cbor2 6.1.3
certifi... 2025.11.12 cffi 2.0.0 charset-normalizer 3.4.4
clique 8.4.2 cloudpickle 3.1.1 colorido 0.5.8
colorlog 6.10.1 comm 0.2.3 tensores comprimidos 0.17.0
contourpy 1.3.3 criptografia 44.0.1 cuda-bindings 13.3.1
Cuda-core 1.0.1 cuda-pathfinder 1.5.6 cuda-python 13.3.1
cuda-tile 1.3.0 Kit de ferramentas CUDA 13.0.2 cycler 0.11.0
agentes do databricks 1.11.0 databricks-sdk 0.122.0 dataclasses-json 0.6.7
datasets 4.8.5 dbus-python 1.3.2 debugpy 1.8.16
decorador 5.2.1 deltalake 1.5.1 Deprecated 1.3.1
depyf 0.20.0 detect-installer 0.1.0 dill 0.4.0
cache de disco 5.6.3 distlib 0.4.0 dm-tree 0.1.10
dnspython 2.8.0 docstring_parser 0.18.0 einops 0.8.2
validador de email 2.3.0 evaluate 0.4.6 executing 2.2.1
Farama-Notificações 0.0.6 fastapi 0.136.3 fastapi-cli 0.0.28
fastapi-cloud-cli 0.22.1 fastar 0.11.0 fastsafetensors 0.3.2
filelock 3.29.0 flash-attn-4 4.0.0b18 flash_attn 2.8.3
flash_attn_3 3.0.0 flashinfer-cubin 0.6.12 flashinfer-python 0.6.12
fonttools 4.60.1 lista de itens congelados 1.8.0 fsspec 2025.9.0
future 1.0.0 gitdb 4.0.11 GitPython 3.1.45
google-api-core 2.30.3 google-auth 2.53.0 google-cloud-core 2.6.0
google-cloud-storage 3.10.1 google-crc32c 1.8.0 google-resumable-media 2.9.0
googleapis-common-protos 1.71.0 Graphviz 0.21 greenlet 3.2.4
grpcio 1.76.0 grpcio-status 1.76.0 ginásio 1.2.2
h11 0.16.0 hf-xet 1.5.1 hf_transfer 0.1.9
httpcore 1.0.9 httplib2 0.20.4 httptools 0.8.0
httpx 0.28.1 httpx-sse 0.4.3 huggingface_hub 1.22.0
humming-kernels 0.1.6 idna 3.11 ijson 3.5.1
importlib_metadata 8.7.0 iniconfig 2.1.0 interegular 0.3.3
ipyflow-core 0.0.227 ipykernel 6.31.0 ipython 9.7.0
ipython_pygments_lexers 1.1.1 ipywidgets 8.1.7 isodate 0.7.2
jedi 0.19.2 Jinja2 3.1.6 jiter 0.16.0
jmespath 1.0.1 joblib 1.5.2 jsonpatch 1.33
jsonpointer 3.1.1 jsonschema 4.25.0 jsonschema-specifications 2025.9.1
jupyter_client 8.6.3 jupyter_core 5.8.1 jupyterlab_widgets 3.0.15
kiwisolver 1.4.8 langchain 1.3.11 langchain-classic 1.0.8
langchain-community 0.4.2 langchain-core 1.4.7 langchain-protocol 0.0.18
langchain-text-splitters 1.1.2 langgraph 1.2.8 langgraph-checkpoint 4.1.1
langgraph-prebuilt 1.1.0 langgraph-sdk 0.4.2 langsmith 0.9.3
lark 1.2.2 launchpadlib 1.11.0 lazr.restfulclient 0.14.6
lazr.uri 1.0.6 lazy_loader 0.4 librosa 0.11.0
liger_kernel 0.8.1 lightgbm 4.6.0 Relâmpago 2.5.5
lightning-utilities 0.15.3 llguidance 1.7.6 llvmlite 0.47.0
lm-format-enforcer 0.11.3 loguru 0.7.3 lz4 4.4.5
Mako 1.3.12 markdown-it-py 2.2.0 MarkupSafe 3.0.2
malvavisco 3.26.2 matplotlib 3.10.6 matplotlib-inline 0.2.1
mccabe 0.7.0 mcp 1.28.1 mdurl 0.1.2
mistral_common 1.11.5 ml_dtypes 0.5.4 mlflow-skinny 3.13.0
mmh3 5.2.1 padrões de contêineres para hospedagem de modelos 0.1.16 mpmath 1.3.0
msal 1.36.0 msal-extensions 1.3.1 msgpack 1.2.1
msgspec 0.21.1 msrest 0.7.1 multidict 6.7.0
multiprocess 0.70.16 mypy-extensions 1.0.0 narvais 2.7.0
nest-asyncio 1.6.0 networkx 3,5 ninja 1.13.0
nodeenv 1.10.0 numba 0.65.0 numpy 2.3.4
nvidia-cublas 13.1.0.3 nvidia-cuda-cccl 13.3.3.4.1 Nvidia-Cuda-CRT 13.3.73
nvidia-cuda-cupti 13.0.85 nvidia-cuda-nvcc 13.2.78 nvidia-cuda-nvrtc 13.0.88
nvidia-cuda-runtime 13.0.96 nvidia-cuda-tileiras 13.2.78 nvidia-cudnn-cu13 9.19.0.56
nvidia-cudnn-frontend 1.25.0 nvidia-cufft 12.0.0.61 nvidia-cufile 1.15.1.6
nvidia-curand 10.4.0.35 nvidia-cusolver 12.0.4.66 NVIDIA-cuSPARSE 12.6.3.3
nvidia-cusparselt-cu13 0.8.0 nvidia-cutlass-dsl 4.5.2 nvidia-cutlass-dsl-libs-base 4.5.2
nvidia-cutlass-dsl-libs-cu13 4.5.2 nvidia-ml-py 13.590.44 nvidia-nccl-cu13 2.28.9
nvidia-nvjitlink 13.0.88 nvidia-nvshmem-cu13 3.4.5 nvidia-nvtx 13.0.85
nvidia-nvvm 13.2.78 oauthlib 3.2.2 omegaconf 2.3.1
openai 2.37.0 openai-harmony 0.0.8 opencensus 0.11.4
opencensus-context 0.1.3 opencv-python 5.0.0.93 opencv_python_headless 5.0.0.93
opentelemetry-api 1.42.1 opentelemetry-exporter-otlp 1.42.1 opentelemetry-exporter-otlp-proto-common 1.42.1
opentelemetry-exporter-otlp-proto-grpc 1.42.1 opentelemetry-exporter-otlp-proto-http 1.42.1 opentelemetry-exporter-prometheus 0,63b1
opentelemetry-proto 1.42.1 opentelemetry-sdk 1.42.1 opentelemetry-semantic-conventions 0,63b1
opentelemetry-semantic-conventions-ai 0.5.1 optuna 4.5.0 optuna-integration 4.5.0
orjson 3.11.9 ormsgpack 1.12.2 outlines_core 0.2.14
packaging 25.0 pandas 2.3.3 parso 0.8.5
partial-json-parser 0.2.1.1.post7 pathspec 0.12.1 patsy 1.0.1
peft 0.19.1 pexpect 4.9.0 pillow 12.0.0
pip 25.0.1 platformdirs 4.5.0 plotly 6.8.0
pluggy 1.5.0 polares 1.42.1 polars-runtime-32 1.42.1
cachorrinho 1.9.0 prometheus-fastapi-instrumentator 8.0.2 prometheus_client 0.21.1
prompt_toolkit 3.0.52 propcache 0.3.1 proto-plus 1.28.0
protobuf 6.33.5 psutil 7.0.0 ptyprocess 0.7.0
pure_eval 0.2.3 py-cpuinfo 9.0.0 py-spy (ferramenta de monitoramento do Python) 0.4.2
pyarrow 21.0.0 pyasn1 0.6.1 pyasn1_modules 0.4.2
pybase64 1.4.3 pyccolo 0.0.83 pycountry 26.2.16
pycparser 2.23 pydantic 2.13.3 pydantic-extra-types 2.11.1
pydantic-settings 2.14.2 pydantic_core 2.46.3 pydot 4.0.0
pyelftools 0.33 pyflakes 3.2.0 Pygments 2.19.2
PyGObject 3.48.2 pyiceberg 0.11.1 PyJWT 2.10.1
pynndescent 0.6.0 pynvml 8.0.1 pyparsing 3.2.5
pyright 1.1.409 pyroaring 1.1.0 pytesseract 0.3.13
pytest 8.4.2 python-dateutil 2.9.0.post0 python-dotenv 1.2.2
python-json-logger 3.2.1 python-multipart 0.0.32 pytokens 0.2.0
pytorch-lightning 2.5.4 pytorch-ranger 0.1.1 pytz 2025.2
PyYAML 6.0.3 pyzmq 27.1.0 quack-kernels 0.5.0
raio 2.56.0 referencing 0.37.0 regex 2025.10.22
requests 2.32.5 requests-oauthlib 2.0.0 requests-toolbelt 1.0.0
rich 14.2.0 ferramenta rica 0.20.1 rignore 0.7.6
rpds-py 0.28.0 s3transfer 0.14.0 safetensors 0.8.0
scikit-learn 1.7.2 scipy 1.16.3 transformadores de frase 5.5.1
sentencepiece 0.2.1 sentry-sdk 2.64.0 setproctitle 1.3.7
setuptools 80.9.0 shap 0.51.0 shellingham 1.5.4
six 1.17.0 segmentação 0.0.8 smart_open 8.0.0
smmap 5.0.0 sniffio 1.3.0 arquivo de som 0.13.1
soxr 1.1.0 SQLAlchemy 2.0.43 sqlparse 0.5.5
sse-starlette 3.4.5 ssh-import-id 5.11 stack-data 0.6.3
starlette 1.3.1 statsmodels 0.14.5 strictyaml 1.7.3
Supervisor 4.3.0 sympy 1.14.0 tabulate 0.9.0
tenacity 9.1.2 tensorboardX 2.6.5 threadpoolctl 3.5.0
tiktoken 0.13.0 tilelang 0.1.9 tokenize_rt 6.2.0
criadores de token 0.22.1 tokenspeed-mla 0.1.2 tokenspeed-triton 3.7.10.post20260531
tomli 2.0.1 lanterna 2.11.0+cu130 torch-optimizer 0.3.0
torch_c_dlpack_ext 0.1.5 torchaudio 2.11.0+cu130 torchmetrics 1.9.0
torchvision 0,26,0+cu130 tornado 6.5.1 tqdm 4.67.1
traitlets 5.14.3 transformadores 5.8.1 Tritão 3.6.0
trl 0.24.0 typer 0.26.8 typing-inspect 0.9.0
inspeção de digitação 0.4.2 typing_extensions 4.15.0 tzdata 2026.2
Ultralíticos 8.4.90 ultralytics-thop 2.0.20 umap-learn 0.5.12
unattended-upgrades 0.1 urllib3 2.5.0 uuid_utils 0.16.2
uvicorn 0.48.0 uvloop 0.22.1 virtualenv 20.35.4
vllm 0.24.0 wadllib 1.3.6 watchfiles 1.2.0
wcwidth 0.2.13 websockets 15.0.1 wheel 0.45.1
whenever 0.7.3 widgetsnbextension 4.0.14 wrapt 1.17.0
xgboost-cu13 3.2.0 xgrammar 0.2.3 xxhash 3.5.0
yarl 1.22.0 z3-solver 4.15.4.0 zipp 3.23.0
zstandard 0.25.0