Understand DSPy: Programming AI Pipelines
Vložit
- čas přidán 6. 05. 2024
- The origin and evolution of DSPy: Programming AI Pipelines introduces the idea, its link to ColBERT v2, retriever models, modular pipeline generation, descriptive programming, the evolution and the use case of DSPy (DSPy == Declarative Self-improving Language Programs, pythonically).
Q answered: Is DSPy only a Prompt Engineering optimization?
Q answered: Is DSPy expensive for my AI pipeline optimization?
Q answered: Can I substitute DSPy with a simple many-shot In-Context Learning prompt?
#airesearch - Věda a technologie
Thanks for the presentation! it helps a lot for the milvus community
Top tier high-level presentation of the amazing DSPy vs ICL, this will help me a lot in my current line of work, keep it up!
You continue to stay one step ahead of me. Thank you so much!
Thanks for the presentation! Any plans to make a demo?
Nice! please explain about ReAct and also having different modules. Like ReAct and ChainOfThoughts together,
Good info, but a bit too much slide-reading and not enough graphs or going over the code. I've been meaning to replace pure RAG in my pipeline with DSPy, but I haven't found any examples of how to actually do this.... and I'm just a bit afraid of touching this can of worms just yet before seeing someone else do the same. :D
What about release some code with your videos as well?
Sure. can you be more specific? What have you been not able to discover?
@@code4AI How you applied DSPY to optimize the prompt for the sample you meantioned in your video? Still struggle to find the first step into that. ;-)