AI Paper Atlas

Research map / Multi-agent systems and games / Human-AI and goal-conditioned

The Decrypto Benchmark for Multi-Agent Reasoning and Theory of Mind

Andrei Lupu, Timon Willi, Jakob Nicolaus Foerster

ICML 2026 · Multi-agent systems and games › Human-AI and goal-conditioned · approach: Theory-of-mind social reasoning · 0 citations (OpenAlex)

Read the paperCheck your idea against 137,217 papers

AI-extracted. The measured effects below were read by AI from the paper's full text (no summary yet). Numbers and claims can be wrong or missing context, so check the paper before you cite or build on it.

Measured effects reported in the paper

What was changedWhereMeasuredEffect
use Claude 3.7 with extended thinking instead of Llama 3.1-70B as decoderDecrypto decoder · Human games, decodermiscommunications: 18.67 -> 12.67Improved
use DeepSeek-R1-Dist-32B instead of GPT-4o as decoderDecrypto decoder · Human games, decodermiscommunications: 17.33 -> 19.67Hurt
fine-tune Llama 3.1-8B on ToM data (ExploreToM-8B)Llama 3.1-8B-Instruct · Human games, interceptorintercepts: 6.33 -> 4.00Hurt
fine-tune Llama 3.1-8B on ToM data (ExploreToM-8B)Llama 3.1-8B-Instruct · Human games, decodergame length: 3.00 -> 2.53Hurt
scale model from Llama 3.1-8B to 70B as decoderLlama 3.1-Instruct · Human games, decodergame length: 3.00 -> 4.70Improved
use reasoning model DeepSeek-R1 instead of Llama 3.1-70B as interceptorDecrypto interceptor · Human games, interceptorwin rate of human team: 46.67% -> 20.00%Improved
increase hint vocabulary size K for word-embedding baselines (ablation)GloVe/Word2Vec baselines · Decrypto cross-playmiscommunicationsHurt
vary system and user prompts (625 variants) (ablation)Llama 3.1-8B and 70B · Decrypto game lengthaverage turnsNo clear effect
emphasize in PT prompt that Eve does not know the secret keywords (ablation)LLM Alice · Perspective takingpredicted Eve intercept rateNo clear effect
use newer reasoning models (Claude 3.7, o1 high) instead of Llama 3.1-70BLLMs · Theory of mind: weak RC and FBaccuracyHurt

Closest papers by meaning

Writing a paper in this area? scime draft is a free LaTeX editor with AI included, real-time co-authors and a check of your references and venue rules before you submit. It also imports OpenAI Prism projects.

Open scime draft

Get the next scime Dispatch and a monthly field update

Free. We send a confirmation link first; every email has a one-click unsubscribe. Privacy