DSPyWeekly Issue No #7

I Taught a Small LLM to Write Fiction. The Results Weren't What I Expected. - Michael

In the rapidly evolving landscape of Large Language Models (LLMs), achieving high-quality, constrained text generation is a common challenge. While powerful models like GPT-5 or Gemini 2.5 Pro set a high bar, what if you could get similar quality from smaller, more efficient models? This post details an experiment where I tried to achieve that. We’ll walk through a hands-on project where I used the DSPy framework to optimize a Gemma-3-1b model for creative story generation with a novel optimization technique called GEPA.

Prompt Optimization for Language Models with DSPy GEPA Authored by: Behrooz Azarkhalili

This notebook demonstrates how to use DSPy’s GEPA (Generalized Error-driven Prompt Augmentation) optimizer to improve language model performance on mathematical reasoning tasks. We’ll work with the NuminaMath-1.5 dataset and show how GEPA can boost accuracy through automated prompt optimization.

Paper: SHIELD: Classifier-Guided Prompting for Robust and Safer LVLMs

SHIELD, a lightweight safety guardrail that integrates a fine-grained taxonomy of harmful content with tailored policies and rule-based interventions. Makes use of DSPy for some parts of the implementation.

DSPyground 0.2.7 is out

With this update, it has now fully evolved into a harness that seamlessly plugs into existing multi turn Agent environments. ( @aisdk based agents to start with.)

Building State-of-the-Art Enterprise Agents 90x Cheaper with Automated Prompt Optimization | Databricks Blog

This Databricks blog post discusses building cost-effective, enterprise-grade AI agents using automated prompt optimization - GEPA, MIPROv2, SIMBA are compared. It introduces Databricks Agent Bricks, a platform that fine-tunes AI models to improve performance and reduce costs. The post highlights how this optimization can make open-source models outperform more expensive proprietary ones. Ultimately, this approach offers a more efficient and scalable solution for enterprise AI applications.

The Alien Artifact: DSPy and the Cargo Cult of LLM Optimization — Data Monger

As a curator it's imp to also share critical assessment by someone. What's flawed in the writers assessment is the lack of understanding that most engineers don't know the ABC of data science, LLM inner workings and for them DSPy lowers the entry level barrier once they have surpassed the basic programmatic integration with LLMs.

DSPy Meetup Tokyo #1

To our readers in Japan, 2025/11/13 will see the very first DSPy meetup in Japan.

Extract structured data from images with DSPy and ax-llm

Joe Maddalone live coding extracting description from book images. Coded in typescript port of DSPy.

はじめてのDSPy - プロンプトエンジニアリングからの脱却 - YouTube

Checked the transcript and there is a short conversation around DSPy. Worth checking for our Japanese readers.

jmanhype/ace-playbook

Self-improving LLM system using Generator-Reflector-Curator pattern for online learning from execution feedback | Language: Python

tom-doerr/dspy-tweet-optimizer

AI-powered tweet optimization tool using DSPy with hill-climbing algorithm | Language: Python | License: MIT License

Archelunch/awesome-dspy-agents

Language: Python

raja-patnaik/dspy-examples

Examples for how to use DSPY and GEPA | Language: Python

weaviate/retrieve-dspy

A collection of Compound Retrieval Systems implemented with DSPy and Weaviate. | Language: Python | License: MIT License

jmanhype/AgentLearningEE

Agent Learning via Early Experience - Bootstrap agent training without reward signals using DSPy | Language: Python

nshkrdotcom/ds_ex

DSPEx - Declarative Self-improving Elixir | A BEAM-Native AI Program Optimization Framework | Language: Elixir | License: MIT License

originalankur/mlflow-dspy-llmops

Accompanying code for Learning End-to-End LLM Lifecycle Management with MLflow talk at Bangalore Python User Group | Language: Python | License: MIT License

ktzsh/dspy-forge

DSPy Forge: A visual tool for building and deploying compound agent programs using DSPy. | Language: Python | License: MIT License

AdminLilypad/DSPY_interpreter