v0.5.0

🚂 UpTrain v0.5 Release Notes 🚀

We're excited to roll out UpTrain v0.5, packed with powerful enhancements to elevate your language model training experience! 🌟

Key Features:

1. Cutting-edge GPT Model Update 🎉

   • UpTrain now leverages the latest and greatest GPT model by default, ensuring top-notch performance and enhanced capabilities.

2. Fine-Tuned Control with Prompt Injection 🤯

   • Introducing Prompt Injection for precise customization! Tailor your prompts to optimize training outcomes and harness the full potential of your models.

3. Response Alignment and Sincerity Evaluations 🧠

   • UpTrain v0.5 introduces Response Alignment and Response Sincerity evaluations aligned with specific scenarios. Achieve even more contextually relevant and sincere responses from your models.

4. Revamped Evaluation Scoring 📈

   • We've overhauled evaluation scoring to provide more accurate insights into your model's performance. Understand your model's strengths and areas for improvement with enhanced clarity.

5. JSON Data Support for Response Format 🚀

   • Now, UpTrain supports JSON data as a response format, offering greater flexibility and compatibility with various applications and systems.

6. Fresh and Improved Prompt Templates 🌈

   • Enjoy an array of updated and improved prompt templates, carefully crafted to spark creativity and produce exceptional training results. Unleash the potential of your models with our refined prompts.

Upgrade to UpTrain v0.5 today and take your language model training to new heights! 🚀🌐 If you have any feedback or suggestions, feel free to reach out. Happy training! 🤖✨

v0.4.9

UpTrain Release Notes v0.4.9 🚀

We're thrilled to unveil the latest UpTrain release (v0.4.9), loaded with enhancements and exciting features to elevate your experience. Let's dive into the details:

Features and Enhancements:

1. 🌐 Improved Azure OpenAI Client for a Seamless Experience

Experience a smoother workflow with our enhanced Azure OpenAI client, making UpTrain integration with your Azure services even more seamless. Tap into the full power of Azure OpenAI effortlessly.

2. 🔗 Up-to-Date OpenAI, OpenAI Evals, and Litellm Support

Stay ahead of the curve! We've updated the UpTrain codebase to ensure compatibility with the latest releases of OpenAI, OpenAI Evals, and Litellm. Embrace cutting-edge features and improvements seamlessly.

3. ✅ New Valid Response Operator

Introducing the Valid Response Operator – a powerful addition to UpTrain! Easily check response validity for any query, ensuring precision and relevance in your interactions. Fine-tune your models with confidence.

4. 🔄 Enhanced JSON and MongoDB Readers

Investing in efficiency! We've refined our JSON and MongoDB readers for a more robust data processing experience. Enjoy improved data handling capabilities, ensuring optimal performance in parsing your data.

5. 📘 Fresh Tutorials to Boost Your Skills

Explore new horizons with our latest tutorials! Whether you're a seasoned user or just starting, our informative guides will help you make the most of UpTrain's features. Elevate your skills with best practices and advanced techniques.

Happy training! 🚀

v0.4.8

🚀 Release Notes - Version 0.4.7 🚀

We're thrilled to introduce the latest updates and enhancements in this release! 🌟

• Updated methods for easier use
• New tutorials for evaluations

v0.4.7

🚀 Release Notes - Version 0.4.7 🚀

We're thrilled to introduce the latest updates and enhancements in this release! 🌟

• New MongoDB Operator lets you integrate with your MongDB Database!
• Improved Conversation Evaluation Tutorial

v0.4.6

[Fix from previous release: Updated the version in pyproject.toml]

🚀 Release Notes - Version 0.4.6 🚀

We're thrilled to introduce the latest updates and enhancements in this release! 🌟

🆕 New Features and Improvements:

We now support multi-turn dialog conversation evaluations. Check out ConversationSatisfaction Eval which assesses if the user is satisfied with the chatbot responses or not. More evals to come in upcoming releases.

Bring your embedding model. You can now call your custom embedding model via an API call inside UpTrain. See the code here.

Llama Index Integration: UpTrain now integrates with Llama Index. We introduce EvalLlamaIndex which allows you to run your LLM pipeline via LlamaIndex and evaluate the responses seamlessly with UpTrain. Checkout the tutorial here

We're committed to making your experience with our platform even better. Thank you for your continued support and feedback! 🙌
For more details and bug fixes, please refer to the full release documentation. Feel free to reach out to us with any questions or feedback. 📧💬

Happy evaluating! 🧪🔬🚀

v0.4.5

🚀 Release Notes - Version 0.4.5 🚀

We're thrilled to introduce the latest updates and enhancements in this release! 🌟

🆕 New Features and Improvements:

1. We now support multi-turn dialog conversation evaluations. Check out ConversationSatisfaction Eval which assesses if the user is satisfied with the chatbot responses or not. More evals to come in upcoming releases.

2. Bring your embedding model. You can now call your custom embedding model via an API call inside UpTrain. See the code here.

3. Llama Index Integration: UpTrain now integrates with Llama Index. We introduce EvalLlamaIndex which allows you to run your LLM pipeline via LlamaIndex and evaluate the responses seamlessly with UpTrain. Checkout the tutorial here

We're committed to making your experience with our platform even better. Thank you for your continued support and feedback! 🙌
For more details and bug fixes, please refer to the full release documentation. Feel free to reach out to us with any questions or feedback. 📧💬

Happy evaluating! 🧪🔬🚀

v0.4.4

Increase timeout while running evaluations on server

v0.4.3

Fixes issues with Python dependencies

v0.4.2

🚀 Release Notes - Version 0.4.2 🚀

We're thrilled to introduce the latest updates and enhancements in this release! 🌟

🆕 New Features and Improvements:

1. We added a new check called Response Matching to quantify the similarity between LLM-generated response and ground truth response. We support three modes:
   a. Exact Match
   b. Words Overlap
   c. LLM-based match
2. Thrilled to announce the beta release of topic clustering where we automatically cluster your input queries and assign topics to them. We are building advanced features to give insights on which topics the model is under-performing - stay tuned1
3. Fix issues with the guideline adherence check

v0.4.1

Release Notes - Version 0.4.1

• Reduce Dependencies
• Add support for Python-3.12