← Back to Blog

Introducing New Dataset Manager and Open-Source Inference

It's now easier than ever to fine-tune and serve open-source models, all on one integrated platform.

Open-source LLM Fine-tuning Platform


Wed Jun 19 2024


Felix Wunderlich



New Release

Today marks a big milestone for FinetuneDB and our community. We’re excited to unveil our latest release that is setting new standards in how you manage datasets and fine-tune open-source LLMs.

Key Updates:

  • Open-Source Model Fine-Tuning and Serving: Easily fine-tune and deploy models like Llama 3 from FinetuneDB, to easily experiment with various models to find the best fit for your use case.
  • Version-Controlled Dataset Manager: Confidently manage and iterate your existing datasets with version control, visual function calling editor, and automatic dataset validation.
  • Seamless Dataset Creation from Production Data: Quickly create high-quality datasets from production data, making the switch from proprietary to open-source models easy.

Introducing New Fine-Tuning Dataset Manager

Our new Dataset Manager brings cutting-edge functionality, enabling you to manage your fine-tuning datasets with ease. Here’s how it transforms your fine-tuning workflow:

Version Controlled Dataset Manager

Keep track of changes and experiment confidently with version histories, making sure you can always revert or compare different dataset versions. Easily edit and organize your data with our user-friendly graphical interface, enabling quicker adjustments and a clearer overview of your datasets. Work together in real-time with non-technical domain experts, share insights, and manage permissions with team members.


Version controlled dataset manager

Version control means reliability in your projects by making sure that any changes or updates can be tracked and reversed if needed.

Automatic Dataset Validation

Our system now automatically validates training data integrity to improve model training outcomes.


Identify issues fast.

This increases the confidence in your data’s quality, which directly contributes to the reliability and performance of your trained models.

Visual Function Calling Editor

Simplify complex function applications with a visual interface, making it easier to apply transformations and data enrichment without code.


Visual function calling editor

This reduces the technical barrier enabling also non-technical team members to contribute more effectively to the dataset creation and fine-tuning process.

Introducing Open Source LLM Fine-Tuning and Serving

In our pursuit to make open-source LLM customization more accessible, we’re excited to announce that our platform now supports open-source fine-tuning and inference, powered by Cloudflare Workers.

  • Flexible Model Selection: Fine-tune, test, and compare different open-source models to find the best fit for your specific use case.
  • Enhance Performance: Use our optimized fine-tuning processes to improve the accuracy and efficiency of your models.
  • Reduce Costs: Cut down on costs by leveraging open-source models like Llama 3 that match or exceed the capabilities of proprietary models.

Flexible model selection

Combining dataset management with direct access to open-source LLM fine-tuning and inference, all in one place, makes it easier than ever to switch between foundation models for cost and performance improvements.

Open-source Inference API Pricing

Llama-8b: $0.4 per million tokens

Llama-v3-70b-instruct: $1.1 per million tokens

Mixtral-8x22b-instruct: $0.8 per million tokens

Mixtral-8x7b-instruct: $0.8 per million tokens

Compared to GPT-4o: $10 per million tokens (combined)

Join Us on This Journey

Whether you’re evaluating a use case or simply curious about our capabilities, reach out to us and discover how FinetuneDB can empower your projects.