Tech Blog | Featured Capital One Technology

Utilizing an alignment-first strategy to transfer advanced reasoning skills to a non-reasoning model (NeurIPS).

An efficient solution to enforce the well-typedness of LLM functions. (EurIPS)

A systematic study of Mixture of Experts (MoE) continual pre-training. (NeurIPS)

A conversational dataset specifically designed to capture and manage inter-tool dependencies across diverse domains. (NeurIPS)

A novel reward modeling framework that is rubric-agnostic, generalizable and provides reasoned score assignments. (NeurIPS)

Visual reasoning models that achieve SoTA performance using an order of magnitude fewer training samples. (NeurIPS)

A study across models and benchmarks and an alternative test-time scaling approach. (NeurIPS)

An RL proxy task that trains VLMs to localize synthetic hallucinations injected into human-written captions. (NeurIPS)

An evaluation architecture that incorporates reasoning and iterative refinement. (NeurIPS)