# Prismatic Synthesis: Gradient-based Data Diversification Boosts Generalization in LLM Reasoning

arXiv: https://arxiv.org/abs/2505.20161 · HTML: https://mlsanity.com/paper/2505.20161/

## Key facts

- Cited by 40 papers in the map (data through 14 Oct 2026).
- Placed in Reasoning benchmarks and evaluation, one of 17 directions in LLM reasoning.
- Builds on 50 of its 72 references that are in the map.
- First posted on arXiv 26 May 2025.

## Placed in

- LLM reasoning / [Reasoning benchmarks and evaluation](https://mlsanity.com/lr/reasoning-benchmarks-and-evaluation/)

## Builds on (most cited first, 30 of 50)

- [GPT-4 Technical Report](https://mlsanity.com/paper/2303.08774/) (2023)
- [The Llama 3 Herd of Models](https://mlsanity.com/paper/2407.21783/) (2024)
- [Training Verifiers to Solve Math Word Problems](https://mlsanity.com/paper/2110.14168/) (2021)
- [DeepSeek-R1: Incentivizing Reasoning Capability in LLMs via Reinforcement Learning](https://mlsanity.com/paper/2501.12948/) (2025)
- [Measuring Mathematical Problem Solving With the MATH Dataset](https://mlsanity.com/paper/2103.03874/) (2021)
- [Qwen2.5 Technical Report](https://mlsanity.com/paper/2412.15115/) (2024)
- [Let's Verify Step by Step](https://mlsanity.com/paper/2305.20050/) (2023)
- [DeBERTa: Decoding-enhanced BERT with Disentangled Attention](https://mlsanity.com/paper/2006.03654/) (2020)
- [Beyond the Imitation Game: Quantifying and extrapolating the capabilities of language models](https://mlsanity.com/paper/2206.04615/) (2022)
- [Solving Quantitative Reasoning Problems with Language Models](https://mlsanity.com/paper/2206.14858/) (2022)
- [s1: Simple test-time scaling](https://mlsanity.com/paper/2501.19393/) (2025)
- [OlympiadBench: A Challenging Benchmark for Promoting AGI with Olympiad-Level Bilingual Multimodal Scientific Problems](https://mlsanity.com/paper/2402.14008/) (2024)
- [LIMA: Less Is More for Alignment](https://mlsanity.com/paper/2305.11206/) (2023)
- [Super-NaturalInstructions: Generalization via Declarative Instructions on 1600+ NLP Tasks](https://mlsanity.com/paper/2204.07705/) (2022)
- [Large Language Models Can Be Easily Distracted by Irrelevant Context](https://mlsanity.com/paper/2302.00093/) (2023)
- [The Flan Collection: Designing Data and Methods for Effective Instruction Tuning](https://mlsanity.com/paper/2301.13688/) (2023)
- [MetaMath: Bootstrap Your Own Mathematical Questions for Large Language Models](https://mlsanity.com/paper/2309.12284/) (2023)
- [Towards General Text Embeddings with Multi-stage Contrastive Learning](https://mlsanity.com/paper/2308.03281/) (2023)
- [MTEB: Massive Text Embedding Benchmark](https://mlsanity.com/paper/2210.07316/) (2022)
- [Deep Batch Active Learning by Diverse, Uncertain Gradient Lower Bounds](https://mlsanity.com/paper/1906.03671/) (2019)
- [Estimating Training Data Influence by Tracing Gradient Descent](https://mlsanity.com/paper/2002.08484/) (2020)
- [Textbooks Are All You Need](https://mlsanity.com/paper/2306.11644/) (2023)
- [Textbooks Are All You Need II: phi-1.5 technical report](https://mlsanity.com/paper/2309.05463/) (2023)
- [LESS: Selecting Influential Data for Targeted Instruction Tuning](https://mlsanity.com/paper/2402.04333/) (2024)
- [LIMO: Less is More for Reasoning](https://mlsanity.com/paper/2502.03387/) (2025)
- [Scaling Synthetic Data Creation with 1,000,000,000 Personas](https://mlsanity.com/paper/2406.20094/) (2024)
- [The Vendi Score: A Diversity Evaluation Metric for Machine Learning](https://mlsanity.com/paper/2210.02410/) (2022)
- [TRAK: Attributing Model Behavior at Scale](https://mlsanity.com/paper/2303.14186/) (2023)
- [GLISTER: Generalization based Data Subset Selection for Efficient and Robust Learning](https://mlsanity.com/paper/2012.10630/) (2020)
- [WANLI: Worker and AI Collaboration for Natural Language Inference Dataset Creation](https://mlsanity.com/paper/2201.05955/) (2022)

## Built on by (most cited first, 30 of 40)

- [Jointly Reinforcing Diversity and Quality in Language Model Generations](https://mlsanity.com/paper/2509.02534/) (2025)
- [Distribution-Aligned Sequence Distillation for Superior Long-CoT Reasoning](https://mlsanity.com/paper/2601.09088/) (2026)
- [The Signal is in the Steps: Local Scoring for Reasoning Data Selection](https://mlsanity.com/paper/2510.03988/) (2025)
- [Backtracking When It Strays: Mitigating Dual Exposure Biases in LLM Reasoning Distillation](https://mlsanity.com/paper/2605.19433/) (2026)
- [DESIGNER: Design-Logic-Guided Multidisciplinary Data Synthesis for LLM Reasoning](https://mlsanity.com/paper/2508.12726/) (2025)
- [Skill-Targeted Adaptive Training](https://mlsanity.com/paper/2510.10023/) (2025)
- [Socratic-MCTS: Test-Time Visual Reasoning by Asking the Right Questions](https://mlsanity.com/paper/2506.08927/) (2025)
- [In Good GRACEs: Principled Teacher Selection for Knowledge Distillation](https://mlsanity.com/paper/2511.02833/) (2025)
- [Think Before You Retrieve: Learning Test-Time Adaptive Search with Small Language Models](https://mlsanity.com/paper/2511.07581/) (2025)
- [Detecting and Suppressing Reward Hacking with Gradient Fingerprints](https://mlsanity.com/paper/2604.16242/) (2026)
- [Privasis: Synthesizing the Largest "Public" Private Dataset from Scratch](https://mlsanity.com/paper/2602.03183/) (2026)
- [MARTI-MARS2: Scaling Multi-Agent Self-Search via Reinforcement Learning for Code Generation](https://mlsanity.com/paper/2602.07848/) (2026)
- [Cram Less to Fit More: Training Data Pruning Improves Memorization of Facts](https://mlsanity.com/paper/2604.08519/) (2026)
- [Long Grounded Thoughts: Synthesizing Visual Problems and Reasoning Chains at Scale](https://mlsanity.com/paper/2511.05705/) (2025)
- [Uncertainty Quantification and Data Efficiency in AI: An Information-Theoretic Perspective](https://mlsanity.com/paper/2512.05267/) (2025)
- [One Sample to Rule Them All: Extreme Data Efficiency in Multidiscipline Reasoning with Reinforcement Learning](https://mlsanity.com/paper/2601.03111/) (2026)
- [Exposing Diversity Bias in Deep Generative Models: Statistical Origins and Correction of Diversity Error](https://mlsanity.com/paper/2602.14682/) (2026)
- [Mechanistic Circuit Identification for Controllable Data Generation](https://mlsanity.com/paper/2608.24065/) (2026)
- [Linguistic and Argument Diversity in Synthetic Data for Function-Calling Agents](https://mlsanity.com/paper/2601.17829/) (2026)
- [Synthetic Data Generation for Training Diversified Commonsense Reasoning Models](https://mlsanity.com/paper/2603.18361/) (2026)
- [ShapE-GRPO: Shapley-Enhanced Reward Allocation for Multi-Candidate LLM Training](https://mlsanity.com/paper/2603.29871/) (2026)
- [MARS2: Scaling Multi-Agent Tree Search via Reinforcement Learning for Code Generation](https://mlsanity.com/paper/2604.14564/) (2026)
- [AcquisitionSynthesis: Targeted Data Generation using Acquisition Functions](https://mlsanity.com/paper/2605.13149/) (2026)
- [DeltaPrompts: Escaping the Zero-Delta Trap in Multimodal Distillation](https://mlsanity.com/paper/2605.15532/) (2026)
- [Which Models Are Our Models Built On? Auditing Invisible Dependencies in Modern LLMs](https://mlsanity.com/paper/2606.12385/) (2026)
- [Forty Shades of Blue: Quality-Diversity Alignment via Mode-Conditioned Reinforcement Learning](https://mlsanity.com/paper/2609.14896/) (2026)
- [The Flaw of Averages: Measuring Benchmark-Level Distributional Robustness](https://mlsanity.com/paper/2509.25671/) (2025)
- [EvoSelect: Data-Efficient LLM Evolution for Targeted Task Adaptation](https://mlsanity.com/paper/2604.26170/) (2026)
- [Evolutionary Task Discovery: Advancing Reasoning Frontiers via Skill Composition and Complexity Scaling](https://mlsanity.com/paper/2605.11666/) (2026)
- [How Much Is a Dataset Worth? Scaling Laws, the Vendi Score, and Matrix Spectral Functions](https://mlsanity.com/paper/2605.29448/) (2026)

## Abstract (arXiv)

Effective generalization in language models depends critically on the diversity of their training data. Yet existing diversity metrics often fall short of this goal, relying on surface-level heuristics that are decoupled from model behavior. This motivates us to ask: What kind of diversity in training data actually drives generalization in language models -- and how can we measure and amplify it? Through large-scale empirical analyses spanning over 300 training runs, carefully controlled for data scale and quality, we show that data diversity can be a strong predictor of generalization in LLM reasoning -- as measured by average model performance on unseen out-of-distribution benchmarks. We introduce G-Vendi, a metric that quantifies diversity via the entropy of model-induced gradients. Despite using a small off-the-shelf proxy model for gradients, G-Vendi consistently outperforms alternative measures, achieving strong correlation (Spearman's ρ ≈ 0.9) with out-of-distribution (OOD) performance on both natural language inference (NLI) and math reasoning tasks. Building on this insight, we present Prismatic Synthesis, a framework for generating diverse synthetic data by targeting underrepresented regions in gradient space. Experimental results show that Prismatic Synthesis consistently improves model performance as we scale synthetic data -- not just on in-distribution test but across unseen, out-of-distribution benchmarks -- significantly outperforming state-of-the-art models that rely on 20 times larger data generator than ours. For example, PrismMath-7B, our model distilled from a 32B LLM, outperforms R1-Distill-Qwen-7B -- the same base model trained on proprietary data generated by 671B R1 -- on 6 out of 7 challenging benchmarks.