AI Paper Atlas

Research map / Large language models / Agents, code and math

Programming by Backprop: Learning Behaviour from Symbolic Descriptions

Jonathan Cook, Silvia Sapora, Arash Ahmadian, Akbir Khan, Tim Rocktäschel, Jakob Nicolaus Foerster, Laura Ruis

ICLR 2026 · Large language models › Agents, code and math · approach: LLM-guided program synthesis

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
add the two-stage proactive curriculum (stage 1 instructions plus executions, then stage 2 evaluation instructions) instead of either stage alone (ablation)Proactive PBB on Qwen3-14B · Random Arithmeticexecution accuracy: negligible-to-low -> 37%Improved
use RL (GRPO) instead of SFT for stage 2 of retroactive PBB (ablation)Retroactive PBB · Random Arithmeticgeneralisation to unpaired programsImproved
express instructions as Python code rather than natural language (ablation)Proactive PBB · Random Arithmeticexecution accuracyImproved
add prompt context augmentation (20 templates) for instructions (ablation)Proactive PBB · Random Arithmeticexecution accuracyImproved
add chain-of-thought execution instead of direct inferenceProactive PBB · Random Arithmeticexecution accuracyImproved
replace up to 100 execution examples with a single instruction in stage 2Proactive PBB vs algorithm distillation · Random Arithmeticexecution fidelity: 1 instruction vs up to 100 examplesNo clear effect
add the proactive curriculum on already familiar programs (ablation)Qwen3 models · Leetcodeexecution accuracyImproved
add retroactive PBB declarative stage on familiar programs (ablation)Retroactive PBB · Leetcodeexecution accuracyNo clear effect
mix PBB cipher instructions with skewed demonstrationsGPT-4o · Ciphersrobustness to shift valuesImproved
add full proactive pipeline versus single stage (ablation)Proactive PBB · Context-free grammarsgrammar complianceImproved
use DPO or GRPO instead of SFT for stage 2 of retroactive PBB (ablation)Retroactive PBB on Llama-3.1-8B · Random ArithmeticgeneralisationImproved
use single-stage SFT mixture instead of two-stage proactive PBB with more repeated code (ablation)Llama-3.1-8B-Instruct · Random Arithmeticunpaired execution accuracyNo clear effect

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