Darwin Family: MRI-Trust-Weighted Evolutionary Merging for Training-Free Scaling of Language-Model Reasoning

• URL: http://arxiv.org/abs/2605.14386v1
• Date: Thu, 14 May 2026 05:09:12 GMT
• Title: Darwin Family: MRI-Trust-Weighted Evolutionary Merging for Training-Free Scaling of Language-Model Reasoning
• Abstract summary: We present a framework for training-free evolutionary merging of large language models via gradient-free weight-space recombination.<n>We ask whether frontier-level reasoning performance can be improved without additional training, by reorganizing latent capabilities already encoded in existing checkpoints.<n>Darwin Family demonstrates that diagnostic-guided evolutionary merging is a practical and reproducible alternative to costly post-training pipelines for reasoning-centric language models.
• Score: 2.0723828938300684
• License: http://creativecommons.org/licenses/by/4.0/
• Abstract: We present Darwin Family, a framework for training-free evolutionary merging of large language models via gradient-free weight-space recombination. We ask whether frontier-level reasoning performance can be improved without additional training, by reorganizing latent capabilities already encoded in existing checkpoints. Darwin introduces three key ideas: (i) a 14-dimensional adaptive merge genome enabling fine-grained component- and block-level recombination; (ii) MRI-Trust Fusion, which adaptively balances diagnostic layer-importance signals with evolutionary search through a learnable trust parameter; and (iii) an Architecture Mapper that enables cross-architecture breeding between heterogeneous model families. Empirically, the flagship Darwin-27B-Opus achieves 86.9% on GPQA Diamond, ranking #6 among 1,252 evaluated models, and outperforming its fully trained foundation model without any gradient-based training. Across scales from 4B to 35B parameters, Darwin models consistently improve over their parents, support recursive multi-generation evolution, and enable a training-free evolutionary merge that combines Transformer- and Mamba-based components. Together, the Darwin Family demonstrates that diagnostic-guided evolutionary merging is a practical and reproducible alternative to costly post-training pipelines for reasoning-centric language models.

Related papers

• Self-Evolving Memory for Generative Recommendation [100.15207958893029]
  Generative recommendation has emerged as a promising end-to-end paradigm for personalized recommendation.<n>Existing evolving strategies, such as continual retraining and distillation-based adaptation, directly update the shared model parameters.<n>We propose a self-evolving memory paradigm for generative recommendation, aiming to enable effective evolution across heterogeneous behavioral patterns.
  arXiv  Detail & Related papers  (2026-09-14T14:00:38Z)
• Evolutionary Discovery of Developmental Reward Schedules in Deep Reinforcement Learning [0.0]
  We propose an evolutionary framework for discovering developmental reward schedules.<n>Agency, novelty, and reactivity are combined through time-varying weights.<n>Our results position evolutionary optimization as a promising approach for developmental reward schedule discovery.
  arXiv  Detail & Related papers  (2026-06-18T18:47:26Z)
• Squeeze Evolve: Unified Multi-Model Orchestration for Verifier-Free Evolution [81.46210789228296]
  We introduce Squeeze Evolve, a unified multi-model orchestration framework for verifier-free evolutionary inference.<n>Our approach is guided by a simple principle: allocate model capability where it has the highest marginal utility.
  arXiv  Detail & Related papers  (2026-04-09T02:14:45Z)
• Guided Self-Evolving LLMs with Minimal Human Supervision [53.111086364268566]
  Unguided self-evolving systems often plateau quickly or degrade as training.<n>R-Few is a guided Self-Play Challenger acquirer framework that incorporates lightweight human oversight through in-context grounding and mixed training.<n>R-Few achieves consistent and iterative improvements across math and general reasoning benchmarks.
  arXiv  Detail & Related papers  (2025-12-02T07:06:11Z)
• Multi-population GAN Training: Analyzing Co-Evolutionary Algorithms [0.3683202928838613]
  Generative adversarial networks (GANs) are powerful generative models but remain challenging to train due to pathologies such as mode collapse and instability.<n>Recent research has explored co-evolutionary approaches, in which populations of generators and discriminators are evolved.<n>This paper presents an empirical analysis of different coevolutionary GAN training strategies, focusing on the impact of selection and replacement mechanisms.
  arXiv  Detail & Related papers  (2025-07-17T14:19:17Z)
• DarwinLM: Evolutionary Structured Pruning of Large Language Models [49.55509443720372]
  Large Language Models (LLMs) have achieved significant success across various NLP tasks.<n>Structured pruning offers an effective solution by compressing models and directly providing end-to-end speed improvements.<n>We propose DarwinLM, a method for training-aware structured pruning.
  arXiv  Detail & Related papers  (2025-02-11T18:59:35Z)
• Evolving Reservoirs for Meta Reinforcement Learning [1.6874375111244329]
  We propose a computational model for studying a mechanism that can enable such a process. At the evolutionary scale, we evolve reservoirs, a family of recurrent neural networks. We employ these evolved reservoirs to facilitate the learning of a behavioral policy through Reinforcement Learning (RL) Our results show that the evolution of reservoirs can improve the learning of diverse challenging tasks.
  arXiv  Detail & Related papers  (2023-12-09T16:11:48Z)
• DARLEI: Deep Accelerated Reinforcement Learning with Evolutionary Intelligence [77.78795329701367]
  We present DARLEI, a framework that combines evolutionary algorithms with parallelized reinforcement learning. We characterize DARLEI's performance under various conditions, revealing factors impacting diversity of evolved morphologies. We hope to extend DARLEI in future work to include interactions between diverse morphologies in richer environments.
  arXiv  Detail & Related papers  (2023-12-08T16:51:10Z)
• eVAE: Evolutionary Variational Autoencoder [40.29009643819948]
  We propose a novel evolutionary variational autoencoder (eVAE) building on the variational information bottleneck (VIB) theory. eVAE integrates a variational genetic algorithm into VAE with variational evolutionary operators including variational mutation, crossover, and evolution. eVAE achieves better reconstruction loss, disentanglement, and generation-inference balance than its competitors.
  arXiv  Detail & Related papers  (2023-01-01T23:54:35Z)
• Epigenetic opportunities for Evolutionary Computation [0.0]
  Evolutionary Computation is a group of biologically inspired algorithms used to solve complex optimisation problems. It can be split into Evolutionary Algorithms, which take inspiration from genetic inheritance, and Swarm Intelligence algorithms, that take inspiration from cultural inheritance. This paper breaks down successful bio-inspired algorithms under a contemporary biological framework based on the Extended Evolutionary Synthesis.
  arXiv  Detail & Related papers  (2021-08-10T09:44:53Z)
• Epigenetic evolution of deep convolutional models [81.21462458089142]
  We build upon a previously proposed neuroevolution framework to evolve deep convolutional models. We propose a convolutional layer layout which allows kernels of different shapes and sizes to coexist within the same layer. The proposed layout enables the size and shape of individual kernels within a convolutional layer to be evolved with a corresponding new mutation operator.
  arXiv  Detail & Related papers  (2021-04-12T12:45:16Z)

This list is automatically generated from the titles and abstracts of the papers in this site.

This site does not guarantee the quality of this site (including all information) and is not responsible for any consequences.