P-tuning v2 is a deep prompt tuning strategy designed as an alternative to full model fine-tuning.
Open source
24 alternativesoumi-ai/oumi
oumi supports fine-tuning, evaluation, and deployment of open-weight language models.
About oumi-ai/oumi
Oumi is an open-source Python framework for fine-tuning, evaluating and deploying open-weight language and vision-language models. Its training support covers supervised fine-tuning, preference methods such as DPO and KTO, and reinforcement learning with GRPO, along with LoRA, QLoRA, full fine-tuning configurations, AWQ quantization and DeepSpeed. Evaluation tooling includes LLM-as-a-judge APIs and a multi-criteria RubricJudge, and the project ships CLI commands such as oumi deploy and oumi analyze plus an MCP server for Claude and Cursor integration. Model support spans families including Qwen, Gemma, Llama, Falcon and gpt-oss. Deployment targets include dedicated inference endpoints on Fireworks and Parasail, and batch APIs cover Anthropic, Fireworks and Together, so some workflows involve external services. Using the project requires a Python environment and its dependencies.
Key features
- Supervised fine-tuning, DPO, KTO and GRPO reinforcement learning for open-weight models
- LoRA, QLoRA and full fine-tuning configurations; AWQ quantization and DeepSpeed support
- Evaluation tooling including LLM-as-a-judge APIs and multi-criteria RubricJudge scoring
- CLI commands for deployment and analysis, plus an MCP server for Claude and Cursor
- Deployment to dedicated inference endpoints on Fireworks and Parasail
- Tool-calling training with executable tool environments and agentic data synthesis
Best for: Machine learning engineers and researchers who want an end-to-end, code-based workflow for fine-tuning, evaluating and deploying open-weight language and vision-language models.
Good for
- Open-source Python project covering training, evaluation and deployment in one stack
- Broad model support across families such as Qwen, Gemma, Llama, Falcon and gpt-oss
- Actively released, with frequent version updates and new training and deployment integrations
Watch out for
- Deployment integrations named in the project are limited to specific providers such as Fireworks and Parasail
- Some inference and batch workflows depend on external services including Anthropic, Fireworks and Together
- Running the framework requires a Python environment and its dependencies
oumi-ai/oumi at a glance
- Website
- github.com/oumi-ai/oumi
- Type
- Open-source project
- Pricing
- Not confirmed. Check the site
- Open source
- Yes · Source code
- Language
- English
- Last checked
- Oct 6, 2026 · Online when checked
- Listed
- Oct 5, 2026
oumi-ai/oumi alternatives
Tools and sites you could use instead, ranked by how closely they match.
PEFT provides parameter-efficient fine-tuning methods for adapting pretrained machine learning models.
Open source
24 alternatives
PyTorch Lightning extension that accelerates and enhances foundation model experimentation with flexible fine-tuning schedules.
Open source
24 alternatives
LightlyTrain supports pretraining, fine-tuning, and distillation for computer vision models.
Open source
24 alternatives
Enables single machine or distributed training and evaluation of deep learning models.
Open source
24 alternatives
autoai automates regression and classification modeling in Python through model search and hyperparameter tuning.
Open source
24 alternativesPopular in ML Training & Optimization
Well-known listings from the same topic.
Edwardedwardlib.orgAerosolveaerosolve.airbnb.techDataturedatature.ioTogether AItogether.aiOptunaoptuna.orgCatBoostcatboost.ai
oumi-ai/oumi FAQ
What kinds of fine-tuning does Oumi support?
The project lists supervised fine-tuning, DPO and KTO preference methods, and GRPO reinforcement learning, with LoRA, QLoRA and full fine-tuning options.
Which models can be trained or evaluated with Oumi?
Documented support includes open-weight families such as Qwen, Qwen3-VL, Gemma, Llama 4, Falcon, InternVL3 and OpenAI's gpt-oss models.
Where can trained models be deployed?
The oumi deploy command supports dedicated inference endpoints on Fireworks and Parasail, and batch API support is listed for Anthropic, Fireworks and Together.
Is something wrong or out of date?
Tell us and an editor will check it.
Own this website?
Claim this listing
Claim the listing and suggest updates yourself. An editor checks every claim first.