#dspy — Public Fediverse posts

The related top HN comment is also worth reading: https://news.ycombinator.com/item?id=47491023

"You're comparing [DSPy] downloads with Langchain, probably the worst package to gain popularity of the last decade. It was just first to market, then after a short while most realized it's horrifically architected, and now it's just coasting on former name recognition while everyone who needs to get shit done uses something lighter like the above two."

Preach! 🙌

#dspy #langchain #hackernews #genai #llms

If you disregard the "DSPy is my favorite hammer and every LLM workflow project is a nail" theme, this blogpost paints a good picture of the natural evolution of LLM engineering at startups with a generative AI product:

https://skylarbpayne.com/posts/dspy-engineering-patterns/

#llms #genai #dspy

🤔 Oh wow, another 11-minute existential crisis about a tool nobody uses! 🚀 DSPy: the "revolutionary" AI that somehow skipped the 'useful' stage and landed straight in the 'why bother' bin. 🤷‍♂️
https://skylarbpayne.com/posts/dspy-engineering-patterns/ #existentialcrisis #AItools #DSPy #innovation #techhumor #whybother #HackerNews #ngated

If Dspy is so great, why isn't anyone using it?

https://skylarbpayne.com/posts/dspy-engineering-patterns/

#HackerNews #Dspy #DspyEngineers #SoftwareTrends #TechDiscussion #UserAdoption

If Dspy is so great, why isn't anyone using it?

https://skylarbpayne.com/posts/dspy-engineering-patterns/

#HackerNews #Dspy #DspyEngineers #SoftwareTrends #TechDiscussion #UserAdoption

If Dspy is so great, why isn't anyone using it?

https://skylarbpayne.com/posts/dspy-engineering-patterns/

#HackerNews #Dspy #DspyEngineers #SoftwareTrends #TechDiscussion #UserAdoption

If Dspy is so great, why isn't anyone using it?

https://skylarbpayne.com/posts/dspy-engineering-patterns/

#HackerNews #Dspy #DspyEngineers #SoftwareTrends #TechDiscussion #UserAdoption

If Dspy is so great, why isn't anyone using it?

https://skylarbpayne.com/posts/dspy-engineering-patterns/

#HackerNews #Dspy #DspyEngineers #SoftwareTrends #TechDiscussion #UserAdoption

Curious how DSPy routes every pipeline step before it touches an LLM? This piece breaks down the gateway class behind DSPy modules and why it matters.

Read More: https://zalt.me/blog/2026/01/dspy-module-gateway

#DSPy #LLM #Python #softwaredesign

If, like me, you weren’t familiar with #DSPy, this was a great talk given at the Databricks Data & AI summit to walk you through why you should care.

If you’re writing prompts by hand in your apps, stop now and read this article by @dbreunig

https://www.dbreunig.com/2025/06/10/let-the-model-write-the-prompt.html

AGI is just around the corner!

I'm learning to use DSPy with GEPA (Genetic-Pareto) prompt optimization. In GEPA a larger "teacher" LLM adjusts the prompt for a smaller "student" LM to perform a specific task as well as possible. The teacher will try many different prompts and evaluate the outcome, in my case the quality of a metadata extraction task.

The larger model (GPT-OSS 120B) just added this to the prompt for the smaller model (Gemma 3 4B):

> Good luck! 🎯

😅

#LLM #LocalLLM #DSPy #GEPA

https://kitfucoda.medium.com/leveling-up-llm-game-development-with-dspy-0b203a27a15f

I explored building a word-guessing game with Large Language Models (LLMs), and found that traditional prompt engineering can be tricky, especially when switching between different LLMs. I re-implemented the game using DSPy, a framework that allows for a more structured, code-driven approach.

The game's logic was broken down into modular components using DSPy, to handle tasks like classifying input and parsing questions. Testing with various local LLMs revealed performance differences in how naturally they understood and responded. DSPy's reward function proved more effective than prompt-based methods for preventing smaller models from revealing the answer.

This project highlights the potential of frameworks like DSPy to create more robust and manageable LLM applications, even with less powerful models. It shows a shift towards more programmatic control in LLM development, emphasizing reusability for future AI solutions.

This hands-on experience underscores the evolving landscape of LLM development. #ArtificialIntelligence #MachineLearning #NLP #GameDevelopment #DSPy #LLM #OpenToWork #GetFediHired #FediHire

#ITByte: #DSPy is the framework for solving advanced tasks with language models (LMs) and retrieval models (RMs).

It unifies techniques for prompting and fine-tuning LMs — and approaches for reasoning, self-improvement, and augmentation with retrieval and tools.

https://knowledgezone.co.in/posts/Intro-to-DSPy-65e1dd544c2922322a9353c9

@alextecplayz @cassidy
- In my usage #DejaVuBackup & #PikaBackups both work very well. #Celeste too for syncing files with #Nextcloud

- Everything #AppManager does is possible w/ #Flatseal, #GnomeSoftware, & opening #flatpak manifests or specific to #Android activities, which are loosely analogous to #DBus services, which #DSpy works well for.

- Never used #Gemini, but #Alpaca works super well with communicating w/ #LLMs via #Ollama. Would make for an awesome #GNOME #SearchProvider

What Is #DSPy? How It Works, Use Cases, and Resources https://www.datacamp.com/blog/dspy-introduction

Storm — an #LLM system that researches a topic (web search) and writes a Wikipedia-styled page

great application of #DSPy!

#LLMs #AI

https://github.com/stanford-oval/storm

Released version 0.0.3 of the Check project:

- Now supports return format Markdown and JSON
- Published a PyPI package `ittia-check` for API connect
- Returns all citations instead of these of the winning verdict only

#FactCheck #FightDisinformation #OpenSource #LLM #RAG #DSPy #search

https://github.com/ittia-research/check/releases/tag/v.0.0.3

Huge thanks to Stanford NLP (DSPy) and Jina AI (Reader) for providing tools and inspiration!

https://github.com/ittia-research/check/releases/tag/v.0.0.2

#FactChecking #FightDisinformation #OpenSource #LLM #RAG #DSPy #search

"The newly introduced compiler in DSPy eliminates any additional prompt engineering or fine-tuning efforts when changing parts in your LM-based applications, such as the LM or data."

#ai #coding #dspy #promptengineering

https://towardsdatascience.com/intro-to-dspy-goodbye-prompting-hello-programming-4ca1c6ce3eb9

Here is a write-up of our project submission for the #GoogleAIHackathon, task was to build a creative app using their #Gemini LLM. We built an LLM (Gemini) based evaluation framework for RAG (Retrieval Augmented Generation) systems optimized with example-driven prompts using #DSPy to generate scores from #RAGAS (-style) metrics. Shoutout to Dave Campbell and Mayank Bhaskar my co-contributors to the project for all their hard work! Links to video and GitHub in post -- https://sujitpal.blogspot.com/2024/05/finetuning-ragas-metrics-using-dspy.html

[ Using Feedback to Improve Your Application: Self Learning GPTs ]
https://blog.langchain.dev/self-learning-gpts/ #LangChain #DSPy #Feedback #LLM

#ITByte: #DSPy is the framework for solving advanced tasks with language models (LMs) and retrieval models (RMs).

It unifies techniques for prompting and fine-tuning LMs — and approaches for reasoning, self-improvement, and augmentation with retrieval and tools.

https://knowledgezone.co.in/posts/Intro-to-DSPy-65e1dd544c2922322a9353c9

I've had a number of discussions with people where some ambiguity comes up between few-shot prompting and fine-tuning. Interestingly, #DSPy abstracts many use cases in a productive way that makes these distinctions irrelevant. It effectively treats the cases as the same, depending on compilation targets.

DSPy is definitely worth checking out!

#LLMs #AI #ml #prompts #machinelearning #gpt4 #gpt35 #AIDev #AIDevelopmentTools

https://github.com/stanfordnlp/dspy