Papersby GlyphCohere research papers
A company publishing mostly on agents and evaluation, best known here for Back to Basics.30 papers · 2,944 citations · latest Sep 2026
Building Multilingual Bridges: Data Mixing as the Pillar of Generalization for In-Language ReasoningOptimized supervised fine-tuning data composition enables reasoning models to consistently process and respond in diverse non-English languages without requiring reasoning supervision in each target language.
Cohere33 upvotesRead →The IOL-AI Challenge: An Open Challenge towards Advancing Linguistic ReasoningAn open competition on unseen Linguistics Olympiad puzzles, graded by the official jury; small 14B systems beat models twice their size.Cohere1 citationsRead →CALIBER: Calibrating Confidence Before and After Reasoning in Language ModelsTrains reasoning models to state their confidence twice, before thinking and after answering, halving calibration error against the best single-estimate method.Cohere1 citationsRead →AI Exposure Scores: what they measure, what they miss, and what comes nextReviews how widely cited scores of which jobs AI can assist are used in policy, what they miss, and newer measures that address it.CohereRead →The Culture Funnel: You Can't Align What isn't in the DataModern LLM pipelines experience a cultural data funnel where explicit cultural signals diminish during post-training, necessitating shifts in training data approaches for better cultural alignment.Cohere1 citationsRead →Soft-SVeRL: Self-Verified Reinforcement Learning with Soft RewardsTurns each prompt into a checklist scored item by item, giving reinforcement learning partial-credit rewards for tasks that cannot be checked automatically.Cohere1 citationsRead →Agents Explore but Agents Ignore: LLMs Lack Environmental CuriosityLLM-based agents fail to exploit discovered unexpected information despite recognizing it, indicating a lack of environmental curiosity that depends on tools, compute, and training data distribution.Cohere2 citations8 upvotesRead →Tiny Aya: Bridging Scale and Multilingual DepthTiny Aya demonstrates high-quality multilingual capabilities with 3.35 billion parameters through region-aware posttraining and balanced language performance.Cohere18 citations10 upvotesRead →CIRCLE: A Framework for Evaluating AI from a Real-World LensA six-stage framework for measuring how deployed AI behaves and affects organisations, linking field tests and red teaming to usable metrics.Cohere1 citationsRead →Unlocking Reasoning Capability on Machine Translation in Large Language ModelsFinds that step-by-step reasoning makes machine translation worse, then trains a model on structured draft-and-revise traces that improve it.Cohere2 citationsRead →SimMerge: Learning to Select Merge Operators from Similarity SignalsPredicts which way of merging language models will work best from cheap similarity signals, avoiding costly merge-and-test searches.Cohere3 citationsRead →The Art of Asking: Multilingual Prompt Optimization for Synthetic DataPrompt-space optimization enhances multilingual large language models by systematically transforming prompts for naturalness, cultural adaptation, and difficulty, leading to improved performance across various metrics.Cohere4 citations1 upvotesRead →EAGER: Entropy-Aware GEneRation for Adaptive Inference-Time ScalingEAGer, a training-free method, uses token-wise entropy to optimize computational resources and improve performance on complex reasoning tasks.Cohere17 citations3 upvotesRead →Making, not Taking, the Best of NFusion-of-N (FusioN) method improves LLM generation quality by synthesizing elements from multiple samples, outperforming Best-of-N in various settings and tasks.Cohere7 citations11 upvotesRead →Verification Limits Code LLM TrainingVerification design and strategies impact code generation performance, showing that richer and more diverse test suites improve capabilities while calibrated verification thresholds enhance data usability.Cohere2 citationsRead →When Life Gives You Samples: The Benefits of Scaling up Inference Compute for Multilingual LLMsThe study examines and proposes new sampling and selection strategies to enhance inference-time compute for multilingual and multi-task large language models, demonstrating significant improvements in win-rates across various languages and tasks.Cohere9 citations11 upvotesRead →Aya Vision: Advancing the Frontier of Multilingual MultimodalityMulti-modal language models are enhanced through synthetic data creation and cross-modal merging techniques, achieving superior performance in multilingual settings.Cohere37 citations14 upvotesRead →Command A: An Enterprise-Ready Large Language ModelCommand A, a multilingual large language model, uses decentralized training with self-refinement and model merging to achieve efficient and top-performing Retrieval Augmented Generation for enterprise use.Cohere70 citations31 upvotesRead →Aya Expanse: Combining Research Breakthroughs for a New Multilingual FrontierA new family of 8B and 32B parameter multilingual language models, Aya Expanse, achieves state-of-the-art performance across multiple languages and outperforms larger models in evaluations.Cohere153 citations9 upvotesRead →Global MMLU: Understanding and Addressing Cultural and Linguistic Biases in Multilingual EvaluationGlobal-MMLU addresses cultural biases and translation artifacts in multilingual datasets by involving professional annotators and offering culturally distinct subsets for more accurate model evaluation.Cohere236 citations21 upvotesRead →M-RewardBench: Evaluating Reward Models in Multilingual SettingsA systematic evaluation of reward models in multilingual settings reveals significant performance gaps and dependencies on translation quality and resource availability.Cohere61 citations10 upvotesRead →To Code, or Not To Code? Exploring Impact of Code in Pre-trainingThe inclusion of code data during pre-training significantly improves LLMs' performance across various natural language and world knowledge tasks, beyond code generation.Cohere61 citations45 upvotesRead →BAM! Just Like That: Simple and Efficient Parameter Upcycling for Mixture of ExpertsBAM enhances the Mixture of Experts framework by fully recycling dense model parameters, including attention layers, leading to better performance and efficiency than previous methods.Cohere20 citations13 upvotesRead →Self-Improving Robust Preference OptimizationSRPO, a self-improving offline RLHF framework, achieves robustness to out-of-distribution tasks by optimizing a min-max objective that jointly enhances self-improvement and generative policies, leading to superior performance compared to DPO.Cohere25 citations19 upvotesRead →Aya 23: Open Weight Releases to Further Multilingual ProgressAya 23 is a powerful multilingual language model for 23 languages, outperforming previous models on a wide range of tasks.Cohere155 citations34 upvotesRead →Replacing Judges with Juries: Evaluating LLM Generations with a Panel of Diverse ModelsEvaluating Large Language Models using a Panel of smaller evaluators outperforms single large evaluators with reduced bias and cost.Cohere418 citations71 upvotesRead →Back to Basics: Revisiting REINFORCE Style Optimization for Learning from Human Feedback in LLMsReinforcement Learning from Human Feedback (RLHF) benefits from simpler REINFORCE-style optimization methods over computationally expensive PPO or "RL-free" methods, enhancing performance and reducing cost in large language models.Cohere804 citations18 upvotesRead →Aya Model: An Instruction Finetuned Open-Access Multilingual Language ModelAya, a multilingual generative language model supporting over 50% lower-resourced languages, excels in both generative and discriminative tasks across 99 languages, and provides extensive evaluations and open-source resources.Cohere438 citations48 upvotesRead →Aya Dataset: An Open-Access Collection for Multilingual Instruction TuningThe initiative builds a human-curated instruction-following dataset spanning 65 languages and creates the largest multilingual collection of instruction-following instances through templating and translating existing datasets across 114 languages, contributing datasets and platforms for participatory research.Cohere253 citations57 upvotesRead →When Less is More: Investigating Data Pruning for Pretraining LLMs at ScalePerplexity outperforms more complex methods for data quality estimation in pruning large language model training datasets, resulting in performance improvement with reduced data.Cohere144 citations17 upvotesRead →