Search results

Dealing with VLM

Status: Open.

#121 In gepa-ai/gepa;

· UlrickBL opened on Oct 22, 2025

Once the score reaches 80%, many minibatches are discarded due to getting perfect scores

Status: Open.

#117 In gepa-ai/gepa;

· LakshyAAAgrawal opened on Oct 19, 2025

Abstract GEPAEngine.run() to have discrete steps that can be run by an external controller

good first issueGood for newcomersGood for newcomers

Status: Open.

#116 In gepa-ai/gepa;

· LakshyAAAgrawal opened on Oct 19, 2025

[Question] Clarification on Reproducing AIME Results from the DSPy Tutorial

Status: Open.

#115 In gepa-ai/gepa;

· Davids048 opened on Oct 18, 2025

Gepa optimize should accept a custom batch_sampler

good first issueGood for newcomersGood for newcomers

Status: Open.

#113 In gepa-ai/gepa;

· CrisRodriguez opened on Oct 15, 2025

GEPA doesn't work in Python 3.9 right now.

Status: Open.

#112 In gepa-ai/gepa;

· LakshyAAAgrawal opened on Oct 13, 2025

GEPA should provide mechanism to provide weights/importance of each data instance.

Status: Open.

#111 In gepa-ai/gepa;

· LakshyAAAgrawal opened on Oct 13, 2025

Do not perform full validation evals for every prompt accepted on the trainset minibatch

Status: Open.

#103 In gepa-ai/gepa;

· LakshyAAAgrawal opened on Oct 9, 2025

Can GEPA accept a teacher module and use that to generate high quality trajectories when the student is unable to?

Status: Open.

#101 In gepa-ai/gepa;

· LakshyAAAgrawal opened on Oct 8, 2025

Reflective dataset contains potentially redundant data for ReAct agent

good first issueGood for newcomersGood for newcomers

Status: Open.

#97 In gepa-ai/gepa;

· maks-ivanov opened on Oct 7, 2025

Can we add new evals?

Status: Open.

#96 In gepa-ai/gepa;

· LakshyAAAgrawal opened on Oct 6, 2025

Estimating number of reflection LM calls

Status: Open.

#95 In gepa-ai/gepa;

· LakshyAAAgrawal opened on Oct 6, 2025