danielmiessler / fabric

fabric is an open-source framework for augmenting humans using AI. It provides a modular framework for solving specific problems using a crowdsourced set of AI prompts that can be used anywhere.
https://danielmiessler.com/p/fabric-origin-story
MIT License
25.52k stars 2.71k forks source link

[Bug]: No CR/LF from streaming response? #1141

Open BryanDollery opened 6 days ago

BryanDollery commented 6 days ago

What happened?

I don't seem to be getting any cr/lf chars in the streaming responses.

I was planning to use streaming and a per-char loop to check if the ai is hallucinating and abort the query if it is.

I get huge blocks of text full of nearly random words, thousands of them - counting chars between cr/lf chars could help mitigate this. But, it's not working because I seem to get no cr/lf chars at all (ascii 10 or 13) in the response. Is this intentional?

Version check

Relevant log output

Here's an example of the output of a summary of a youtube transcription: 

## ONE SENTENCE SUMMARY:Accelerated computing, driven by GPU advancements and machine learning, is transforming industries with enhanced computational power and AI-driven capabilities.## MAIN POINTS:1. Moore's Law no longer drives CPU improvements; accelerated computing is now essential for progress.2. Accelerated computing involves reinventing the computing stack from algorithms to architecture.3. GPUs, like NVIDIA's CUDA, have revolutionized industries such as computer graphics and AI.4. Software development has shifted from traditional coding to machine learning approaches.5. Large-scale data analysis enables translating and understanding various data modalities.6. NVIDIA's Blackwell system exemplifies advanced GPU technology for massive data processing.7. AI model training and inference have new scaling laws, enhancing computational efficiency.8. AI Enterprise and Omniverse platforms empower agents and physical AI applications.9. The demand for AI and foundational models is rapidly increasing across industries.10. Physical AI requires advanced computing systems for training, simulation, and real-world deployment.## TAKEAWAYS:1. The end of Moore's Law emphasizes the need for accelerated computing to enhance performance.2. GPUs have become crucial for advancing AI and computational tasks across multiple domains.3. Machine learning has replaced traditional coding, driving software evolution.4. Blackwell and related technologies enable unprecedented data processing and AI capabilities.5. AI platforms like NVIDIA AI Enterprise and Omniverse are pivotal for future technological growth.

Relevant screenshots (optional)

No response