Skip to main content

MindStudio Service Profiler Advisor

Project description

MindStudio Service Profiling Advisor

介绍

  • 基本功能 根据当前的 benchmark 输出结果以及 service 的 config.json 配置,分析提高 TFTT / Throughput 等的优化点

  • 安装

    pip install msservice-advisor
    
  • 执行 mindie benchmark,输出结果的 instance 文件夹

  • 执行调参建议

    msservice_advisor -i ../instance/
    # msservice_advisor_logger - INFO - <think>
    # msservice_advisor_logger - INFO - num_mem_size_checker
    # msservice_advisor_logger - INFO - check_input_tokens
    # ...
    # msservice_advisor_logger - INFO - find_best_batch_size
    # msservice_advisor_logger - INFO - ==decode==
    # msservice_advisor_logger - INFO - ==prefill==
    # msservice_advisor_logger - INFO - </think>
    # msservice_advisor_logger - INFO -
    # msservice_advisor_logger - INFO - <answer>
    # msservice_advisor_logger - INFO - [config] maxBatchSize
    # msservice_advisor_logger - INFO - [action] set bigger
    # msservice_advisor_logger - INFO - [reason] 目前batch样本太小,建议调大点试试
    # msservice_advisor_logger - INFO -
    # msservice_advisor_logger - INFO - [config] maxPrefillBatchSize
    # msservice_advisor_logger - INFO - [action] set bigger
    # msservice_advisor_logger - INFO - [reason] 目前batch样本太小,建议调大点试试
    # msservice_advisor_logger - INFO -
    # msservice_advisor_logger - INFO - </answer>
    
  • 参数

    参数 说明
    -i, --instance_path benchamrk 输出的 instance 路径
    -s, --service_config_path mindie service config.json path, 默认值为环境变量中的 MIES_INSTALL_PATH
    -t, --target 调参指标, 可选值:ttft, firsttokentime, throughput
    -m, --target_metrics 调参指标的具体项,可选值:average,max,min,P75,P90,SLO_P90,P99,N
    --show 指定是否显示拟合曲线的 matplotlib 绘图
    -l, --log_level 日志级别,可选值 debug, info, warning, error

Project details


Download files

Download the file for your platform. If you're not sure which to choose, learn more about installing packages.

Source Distribution

msservice_advisor-0.0.1a7.tar.gz (10.7 kB view details)

Uploaded Source

Built Distribution

If you're not sure about the file name format, learn more about wheel file names.

msservice_advisor-0.0.1a7-py3-none-any.whl (13.9 kB view details)

Uploaded Python 3

File details

Details for the file msservice_advisor-0.0.1a7.tar.gz.

File metadata

  • Download URL: msservice_advisor-0.0.1a7.tar.gz
  • Upload date:
  • Size: 10.7 kB
  • Tags: Source
  • Uploaded using Trusted Publishing? No
  • Uploaded via: twine/5.0.0 CPython/3.8.19

File hashes

Hashes for msservice_advisor-0.0.1a7.tar.gz
Algorithm Hash digest
SHA256 f90ddafff416cc405450984268ee85b0bf4c82bdcf64098db3f40b36006c481a
MD5 34c3f0fa7a05b7a389f3972a02cc2295
BLAKE2b-256 9e78f254f42910d9110206e6558c760f6b2a15dd76cb426c0211e8dbd52958a5

See more details on using hashes here.

File details

Details for the file msservice_advisor-0.0.1a7-py3-none-any.whl.

File metadata

File hashes

Hashes for msservice_advisor-0.0.1a7-py3-none-any.whl
Algorithm Hash digest
SHA256 72e60f7b78b044dd96f50220ba1a4fc2f6d97f9380c788961043fb6732133b5c
MD5 fdd673337c9272e6b41c0a7241f24e07
BLAKE2b-256 774d3a799134e4c9df4573e4da5eb77e7f329fed14cf9811cd5cedf6058f22e8

See more details on using hashes here.

Supported by

AWS Cloud computing and Security Sponsor Datadog Monitoring Depot Continuous Integration Fastly CDN Google Download Analytics Pingdom Monitoring Sentry Error logging StatusPage Status page