Back to Details

together-fine-tuning

Compare original and translation side by side

🇺🇸

Original

English
🇨🇳

Translation

Chinese

Together Fine-Tuning

Together 微调服务

Overview

概述

Use Together AI fine-tuning when the user needs to adapt a model to their own data or behavior.
Supported workflows in this repo:
  • LoRA fine-tuning
  • full fine-tuning
  • DPO preference tuning
  • VLM fine-tuning
  • function-calling fine-tuning
  • reasoning fine-tuning
  • BYOM upload paths
当用户需要根据自有数据或业务行为适配模型时，可使用Together AI微调功能。
本仓库支持的工作流如下：
  • LoRA 微调
  • 全参数微调
  • DPO 偏好调优
  • VLM 微调
  • 函数调用微调
  • 推理能力微调
  • BYOM 上传路径

When This Skill Wins

适用场景

  • Train a model on custom instruction or conversational data
  • Improve function-calling reliability with supervised examples
  • Train on preferences rather than only demonstrations
  • Fine-tune multimodal or reasoning-oriented models
  • Deploy a fine-tuned output model later through dedicated endpoints
  • 基于自定义指令或对话数据训练模型
  • 通过有标注示例提升函数调用的可靠性
  • 基于偏好数据而非仅演示数据训练模型
  • 微调多模态或面向推理能力的模型
  • 后续可通过专属端点部署微调后的输出模型

Hand Off To Another Skill

跳转其他功能

  • Use 
    together-chat-completions
    for plain inference without training
  • Use 
    together-evaluations
    to measure a model before or after tuning
  • Use 
    together-dedicated-endpoints
    to host the resulting tuned model
  • Use 
    together-gpu-clusters
    only when the user needs raw infrastructure rather than managed tuning
  • 无需训练仅需基础推理时，使用 
    together-chat-completions
  • 需要在调优前后评估模型效果时，使用 
    together-evaluations
  • 需要托管调优后生成的模型时，使用 
    together-dedicated-endpoints
  • 仅当用户需要原始基础设施而非托管微调服务时，使用 
    together-gpu-clusters

Quick Routing

快速路径指引

  • Standard LoRA or full fine-tuning
    • Start with scripts/finetune_workflow.py
    • Read references/data-formats.md
  • DPO preference tuning
    • Start with scripts/dpo_workflow.py
  • Function-calling tuning
    • Start with scripts/function_calling_finetune.py
  • Reasoning tuning
    • Start with scripts/reasoning_finetune.py
  • VLM tuning
    • Start with scripts/vlm_finetune.py
  • Model support and deployment options
    • Read references/supported-models.md
    • Read references/deployment.md
  • 标准LoRA或全参数微调
    • 从 scripts/finetune_workflow.py 开始
    • 阅读 references/data-formats.md
  • DPO偏好调优
    • 从 scripts/dpo_workflow.py 开始
  • 函数调用调优
    • 从 scripts/function_calling_finetune.py 开始
  • 推理能力调优
    • 从 scripts/reasoning_finetune.py 开始
  • VLM调优
    • 从 scripts/vlm_finetune.py 开始
  • 模型支持与部署选项
    • 阅读 references/supported-models.md
    • 阅读 references/deployment.md

Workflow

工作流程

  1. Choose the tuning method that matches the desired behavior change.
  2. Validate dataset format before spending tokens on training.
  3. Upload training data and keep the returned file ID.
  4. Create the job with explicit method-specific parameters.
  5. Monitor job state, events, and checkpoints before handing off to deployment.
  1. 选择符合预期能力调整需求的调优方法。
  2. 在投入算力进行训练前验证数据集格式。
  3. 上传训练数据并留存返回的文件ID。
  4. 使用对应方法的明确参数创建训练任务。
  5. 在跳转至部署环节前，监控任务状态、事件和检查点。

High-Signal Rules

重要注意事项

  • Python scripts require the Together v2 SDK (
    together>=2.0.0
    ). If the user is on an older version, they must upgrade first: 
    uv pip install --upgrade "together>=2.0.0"
    .
  • Prefer LoRA unless the user has a specific reason to pay for full fine-tuning.
  • Keep data-format validation close to the upload step so bad files fail early.
  • Treat deployment as a separate phase; fine-tuning success does not automatically mean serving success.
  • Use the method-specific script instead of overloading one generic workflow for all modes.
  • Parameterize dataset paths, model IDs, and suffixes in automation instead of embedding one demo dataset forever.
  • Python脚本要求使用Together v2 SDK（
    together>=2.0.0
    ）。如果用户使用的是旧版本，必须先升级：
    uv pip install --upgrade "together>=2.0.0"
    。
  • 优先选择LoRA，除非用户有明确理由付费使用全参数微调。
  • 在上传步骤前就近完成数据格式验证，以便异常文件尽早报错。
  • 将部署视为独立阶段；微调成功并不自动意味着部署可用。
  • 使用对应方法的专用脚本，而非复用一套通用工作流覆盖所有模式。
  • 在自动化流程中对数据集路径、模型ID和后缀做参数化处理，不要固定嵌入单一演示数据集。

Resource Map

资源索引

  • Data formats: references/data-formats.md
  • Supported models: references/supported-models.md
  • Deployment guide: references/deployment.md
  • LoRA or full workflow: scripts/finetune_workflow.py
  • DPO workflow: scripts/dpo_workflow.py
  • Function-calling workflow: scripts/function_calling_finetune.py
  • Reasoning workflow: scripts/reasoning_finetune.py
  • VLM workflow: scripts/vlm_finetune.py
  • 数据格式: references/data-formats.md
  • 支持的模型: references/supported-models.md
  • 部署指南: references/deployment.md
  • LoRA或全参数工作流: scripts/finetune_workflow.py
  • DPO工作流: scripts/dpo_workflow.py
  • 函数调用工作流: scripts/function_calling_finetune.py
  • 推理能力工作流: scripts/reasoning_finetune.py
  • VLM工作流: scripts/vlm_finetune.py

Official Docs

官方文档