AI
EpiCon: Collective Agent Learning through Co-Evolving Multimodal Memory
Publication: Hugging Face Daily Papers
Author: Not available in collected metadata
Source published: 28/09/2026, 20:00:00 UTC
Briefing created: 30/09/2026, 05:22:47 UTC
Based on a feed excerpt; full article content has not been acquired. Source evidence retrieved 30/09/2026, 05:18:24 UTC.
Briefings
EpiCon is a framework for collective agent learning using shared multimodal memory without host model updates. It employs two independently trained 2B models: a memory controller and a tree self-organizer. The controller refines textual guidance and visual evidence across attempts, while the self-organizer consolidates lessons hierarchically. Evaluated on eleven benchmarks across four multimodal domains, the framework improves macro-average scores by 1.7 to 4.9 points over No Memory and reduces memory-operation time by 67% to 74% relative to backbone-sized models. A frozen memory bank improves other systems even with a single solving attempt.
Key points
- EpiCon uses two independently trained 2B models: a memory controller and a tree self-organizer to link question-level memory evolution to a persistent experience bank.
- The framework improves macro-average scores by 1.7 to 4.9 points over No Memory across eleven benchmarks spanning four multimodal task domains.
Context and limitations
Evaluation was conducted using two harnesses and multiple backbones; specific results for single-agent performance with a frozen bank are noted but not detailed in the evidence.
Generated by Bittrees-hosted qwen35-2b sha256:aaf42c8b7c3c on 30/09/2026, 05:22:47 UTC. Based on publisher-provided feed evidence; may not cover the full original article. Check the source for details.
View archived briefing on IPFS
bafybeiby6dddkf5ltseekh7t44joy5upfpdut6manavv3c5oqpzl7x4rli