[Usage]: vllm OpenAI API Offline Batch Inference

[pesc101]

Your current environment

2024-09-18 13:46:38.125558: E external/local_xla/xla/stream_executor/cuda/cuda_fft.cc:485] Unable to register cuFFT factory: Attempting to register factory for plugin cuFFT when one has already been registered
2024-09-18 13:46:38.139942: E external/local_xla/xla/stream_executor/cuda/cuda_dnn.cc:8454] Unable to register cuDNN factory: Attempting to register factory for plugin cuDNN when one has already been registered
2024-09-18 13:46:38.144110: E external/local_xla/xla/stream_executor/cuda/cuda_blas.cc:1452] Unable to register cuBLAS factory: Attempting to register factory for plugin cuBLAS when one has already been registered
2024-09-18 13:46:38.156547: I tensorflow/core/platform/cpu_feature_guard.cc:210] This TensorFlow binary is optimized to use available CPU instructions in performance-critical operations.
To enable the following instructions: AVX2 FMA, in other operations, rebuild TensorFlow with the appropriate compiler flags.
2024-09-18 13:46:41.410252: W tensorflow/compiler/tf2tensorrt/utils/py_utils.cc:38] TF-TRT Warning: Could not find TensorRT
PyTorch version: 2.4.0+cu121
Is debug build: False
CUDA used to build PyTorch: 12.1
ROCM used to build PyTorch: N/A

OS: Ubuntu 22.04.4 LTS (x86_64)
GCC version: (Ubuntu 11.4.0-1ubuntu1~22.04) 11.4.0
Clang version: Could not collect
CMake version: version 3.22.1
Libc version: glibc-2.35

Python version: 3.12.5 | packaged by conda-forge | (main, Aug  8 2024, 18:36:51) [GCC 12.4.0] (64-bit runtime)
Python platform: Linux-5.15.0-112-generic-x86_64-with-glibc2.35
Is CUDA available: False
CUDA runtime version: Could not collect
CUDA_MODULE_LOADING set to: N/A
GPU models and configuration: 
GPU 0: NVIDIA RTX A6000
GPU 1: NVIDIA RTX A6000
GPU 2: NVIDIA RTX A6000
GPU 3: NVIDIA A100 80GB PCIe
GPU 4: NVIDIA RTX A6000

Nvidia driver version: 550.90.07
cuDNN version: Could not collect
HIP runtime version: N/A
MIOpen runtime version: N/A
Is XNNPACK available: True

CPU:
Architecture:                       x86_64
CPU op-mode(s):                     32-bit, 64-bit
Address sizes:                      48 bits physical, 48 bits virtual
Byte Order:                         Little Endian
CPU(s):                             128
On-line CPU(s) list:                0-127
Vendor ID:                          AuthenticAMD
Model name:                         AMD EPYC 7543 32-Core Processor
CPU family:                         25
Model:                              1
Thread(s) per core:                 2
Core(s) per socket:                 32
Socket(s):                          2
Stepping:                           1
Frequency boost:                    enabled
CPU max MHz:                        3737.8899
CPU min MHz:                        1500.0000
BogoMIPS:                           5599.92
Flags:                              fpu vme de pse tsc msr pae mce cx8 apic sep mtrr pge mca cmov pat pse36 clflush mmx fxsr sse sse2 ht syscall nx mmxext fxsr_opt pdpe1gb rdtscp lm constant_tsc rep_good nopl nonstop_tsc cpuid extd_apicid aperfmperf rapl pni pclmulqdq monitor ssse3 fma cx16 pcid sse4_1 sse4_2 movbe popcnt aes xsave avx f16c rdrand lahf_lm cmp_legacy svm extapic cr8_legacy abm sse4a misalignsse 3dnowprefetch osvw ibs skinit wdt tce topoext perfctr_core perfctr_nb bpext perfctr_llc mwaitx cpb cat_l3 cdp_l3 invpcid_single hw_pstate ssbd mba ibrs ibpb stibp vmmcall fsgsbase bmi1 avx2 smep bmi2 invpcid cqm rdt_a rdseed adx smap clflushopt clwb sha_ni xsaveopt xsavec xgetbv1 xsaves cqm_llc cqm_occup_llc cqm_mbm_total cqm_mbm_local clzero irperf xsaveerptr rdpru wbnoinvd amd_ppin arat npt lbrv svm_lock nrip_save tsc_scale vmcb_clean flushbyasid decodeassists pausefilter pfthreshold v_vmsave_vmload vgif v_spec_ctrl umip pku ospke vaes vpclmulqdq rdpid overflow_recov succor smca
Virtualization:                     AMD-V
L1d cache:                          2 MiB (64 instances)
L1i cache:                          2 MiB (64 instances)
L2 cache:                           32 MiB (64 instances)
L3 cache:                           512 MiB (16 instances)
NUMA node(s):                       2
NUMA node0 CPU(s):                  0-31,64-95
NUMA node1 CPU(s):                  32-63,96-127
Vulnerability Gather data sampling: Not affected
Vulnerability Itlb multihit:        Not affected
Vulnerability L1tf:                 Not affected
Vulnerability Mds:                  Not affected
Vulnerability Meltdown:             Not affected
Vulnerability Mmio stale data:      Not affected
Vulnerability Retbleed:             Not affected
Vulnerability Spec rstack overflow: Mitigation; safe RET
Vulnerability Spec store bypass:    Mitigation; Speculative Store Bypass disabled via prctl and seccomp
Vulnerability Spectre v1:           Mitigation; usercopy/swapgs barriers and __user pointer sanitization
Vulnerability Spectre v2:           Mitigation; Retpolines; IBPB conditional; IBRS_FW; STIBP always-on; RSB filling; PBRSB-eIBRS Not affected; BHI Not affected
Vulnerability Srbds:                Not affected
Vulnerability Tsx async abort:      Not affected

Versions of relevant libraries:
[pip3] numpy==1.26.4
[pip3] nvidia-cublas-cu12==12.1.3.1
[pip3] nvidia-cuda-cupti-cu12==12.1.105
[pip3] nvidia-cuda-nvrtc-cu12==12.1.105
[pip3] nvidia-cuda-runtime-cu12==12.1.105
[pip3] nvidia-cudnn-cu12==9.1.0.70
[pip3] nvidia-cufft-cu12==11.0.2.54
[pip3] nvidia-curand-cu12==10.3.2.106
[pip3] nvidia-cusolver-cu12==11.4.5.107
[pip3] nvidia-cusparse-cu12==12.1.0.106
[pip3] nvidia-ml-py==12.560.30
[pip3] nvidia-nccl-cu12==2.20.5
[pip3] nvidia-nvjitlink-cu12==12.6.68
[pip3] nvidia-nvtx-cu12==12.1.105
[pip3] optree==0.12.1
[pip3] pytorch-gpu==0.0.1
[pip3] pyzmq==26.2.0
[pip3] torch==2.4.0
[pip3] torchvision==0.19.0
[pip3] transformers==4.44.2
[pip3] triton==3.0.0
[conda] numpy                     1.26.4                   pypi_0    pypi
[conda] nvidia-cublas-cu12        12.1.3.1                 pypi_0    pypi
[conda] nvidia-cuda-cupti-cu12    12.1.105                 pypi_0    pypi
[conda] nvidia-cuda-nvrtc-cu12    12.1.105                 pypi_0    pypi
[conda] nvidia-cuda-runtime-cu12  12.1.105                 pypi_0    pypi
[conda] nvidia-cudnn-cu12         9.1.0.70                 pypi_0    pypi
[conda] nvidia-cufft-cu12         11.0.2.54                pypi_0    pypi
[conda] nvidia-curand-cu12        10.3.2.106               pypi_0    pypi
[conda] nvidia-cusolver-cu12      11.4.5.107               pypi_0    pypi
[conda] nvidia-cusparse-cu12      12.1.0.106               pypi_0    pypi
[conda] nvidia-ml-py              12.560.30                pypi_0    pypi
[conda] nvidia-nccl-cu12          2.20.5                   pypi_0    pypi
[conda] nvidia-nvjitlink-cu12     12.6.68                  pypi_0    pypi
[conda] nvidia-nvtx-cu12          12.1.105                 pypi_0    pypi
[conda] optree                    0.12.1                   pypi_0    pypi
[conda] pytorch-gpu               0.0.1                    pypi_0    pypi
[conda] pyzmq                     26.2.0          py312hbf22597_2    conda-forge
[conda] torch                     2.4.0                    pypi_0    pypi
[conda] torchvision               0.19.0                   pypi_0    pypi
[conda] transformers              4.44.2                   pypi_0    pypi
[conda] triton                    3.0.0                    pypi_0    pypi
ROCM Version: Could not collect
Neuron SDK Version: N/A
vLLM Version: 0.6.1.post2@9ba0817ff1eb514f51cc6de9cb8e16c98d6ee44f
vLLM Build Flags:
CUDA Archs: Not Set; ROCm: Disabled; Neuron: Disabled
GPU Topology:
GPU0    GPU1    GPU2    GPU3    GPU4    CPU Affinity    NUMA Affinity   GPU NUMA ID
GPU0     X      NODE    SYS     SYS     SYS     0-31,64-95      0               N/A
GPU1    NODE     X      SYS     SYS     SYS     0-31,64-95      0               N/A
GPU2    SYS     SYS      X      NODE    NODE    32-63,96-127    1               N/A
GPU3    SYS     SYS     NODE     X      NODE    32-63,96-127    1               N/A
GPU4    SYS     SYS     NODE    NODE     X      32-63,96-127    1               N/A

Legend:

  X    = Self
  SYS  = Connection traversing PCIe as well as the SMP interconnect between NUMA nodes (e.g., QPI/UPI)
  NODE = Connection traversing PCIe as well as the interconnect between PCIe Host Bridges within a NUMA node
  PHB  = Connection traversing PCIe as well as a PCIe Host Bridge (typically the CPU)
  PXB  = Connection traversing multiple PCIe bridges (without traversing the PCIe Host Bridge)
  PIX  = Connection traversing at most a single PCIe bridge
  NV#  = Connection traversing a bonded set of # NVLinks

How would you like to use vllm

I want to use the OpenAI library to do offline inference on my local vllm Model.
I use this compose.yml to create an api-server using vllm.

services:
  vllm:
    image: vllm/vllm-openai:latest
    runtime: nvidia
    deploy:
      resources:
        reservations:
          devices:
            - driver: nvidia
              device_ids: ["4"]
              capabilities: [gpu]
    environment:
      - HUGGING_FACE_HUB_TOKEN=${HUGGING_FACE_HUB_TOKEN}
      - MODEL_NAME=${MODEL_NAME}
    ports:
      - "8000:8000"
    volumes:
      - ~/.cache/huggingface:/root/.cache/huggingface
    ipc: "host"
    command: --model ${MODEL_NAME}

When I try to use batch API endpoint like this, I get for both create calls NotFoundError: Error code: 404 - {'detail': 'Not Found'}. The test.jsonl file has the same format as in this tutorial: https://platform.openai.com/docs/guides/batch/getting-started?lang=curl, except that the name of the model is aligned with the correct model name.
I would assume that there is a problem with the endpoints using vllm as backend. Is it possible to use them or initialize them?

client = OpenAI(api_key="EMPTY", base_url="http://localhost:8000/v1")
batch_input_file = client.files.create(
  file=open("test.jsonl", "rb"),
  purpose="batch"
)
client.batches.create(
    input_file_id= batch_input_file.id,
    endpoint="/v1/chat/completions",
    completion_window="24h",
    metadata={
      "description": "nightly eval job"
    }
)

Before submitting a new issue...

• Make sure you already searched for relevant issues, and asked the chatbot living at the bottom right corner of the documentation page, which can answer lots of frequently asked questions.

[DarkLight1337]

You are trying to conduct batch inference using the OpenAI client which connects to the online server.

For offline batch inference via OpenAI API, you should instead use vllm/entrypoints/openai/run_batch.py.

[pesc101]

Okay, but how can I integrate that with my docker compose, because the docker image always uses api_server.py. So I have to create a new docker image for that or does vllm provides one?

[DarkLight1337]

A simpler way would be to run the vLLM docker container as is, then open a new interactive shell inside it and run any commands you want.

[pesc101]

Well, I have tried to run it inside of the docker and it is not working properly.
The run_batch.py endpoint will start a new model instance and will not use the existing api_server and that is not useful.
Also, that is very far away from an integration into the OpenAI Python SDK because I do not want to execute something inside of the docker all the time

[DarkLight1337]

The run_batch.py endpoint will start a new model instance and will not use the existing api_server and that is not useful.
Also, that is very far away from an integration into the OpenAI Python SDK because I do not want to execute something inside of the docker all the time

Yes, this is why it is called offline inference. Feel free to open an issue to request for online support.

[pesc101]

Sure,
I will open an issue for that.
Thanks for your support!

[DarkLight1337]

If you intend to only run batch inference inside Docker, what you can do is modify the image to run something like sleep infinity instead of vllm serve so that no server is started while keeping the container alive, then run offline batch inference manually.

[oswardlx]

@DarkLight1337 I would like to know how the "Offline Inference with the OpenAI Batch file" model of inference in the industry is generally service-oriented or what suggestions are made. Is it to write an HTTP service and then execute Python commands on the HTTP service server?

[DarkLight1337]

Personally I haven't used it. Perhaps you can refer to the original issue #4777