arxiv arXiv cs.CL · 1d ago · research

Fathom-Vaidya improves medical reasoning with rubric-based rewards

from English

The authors introduce Fathom-Vaidya, a 30B parameter model that uses synthetic data and rubric-based reinforcement learning to enhance both diagnostic and clinical healthcare reasoning.

The framework employs rule- and rubric-guided RL on MedBullets-derived questions to improve diagnostic accuracy.
It generates 5.3k synthetic multi-turn scenarios with multi-dimensional rubrics to assess interactive clinical dialogue.
The model achieves over 10% improvement on MedXpertQA and 50.1% accuracy on HealthBench-Hard, surpassing proprietary baselines like GPT-5 (thinking).

This approach demonstrates that targeted synthetic datasets and rubric-based training can systematically improve medical LLM performance across complex diagnostic and conversational tasks.

Importance 2/3 Beats a top-lab benchmark arXiv cs.CL Research paper Training methods

Benchmarks

Benchmark	Model	Score
HealthBench	Fathom-Vaidya	50.1%

Read original

arxiv arXiv cs.LG · 1d ago · 9 views

Fathom-Vaidya improves medical reasoning with rubric-based rewards

The authors introduce Fathom-Vaidya, a 30B parameter model that uses synthetic data and rubric-based reinforcement learning to enhance diagnostic and clinical healthcare reasoning. The training framework first applies rule-guided RL to MedBullets-derived questions for diagnosis, then utilizes 5.3k synthetic multi-turn scenarios with multi-dimensional rubrics for interactive clinical tasks.

arxiv arXiv cs.AI · 1d ago · 12 views

Fathom-Vaidya improves medical reasoning with rubric-based rewards

The authors introduce Fathom-Vaidya, a 30B parameter model that uses synthetic data and rubric-based reinforcement learning to enhance both diagnostic and clinical healthcare reasoning. The training framework first targets diagnostic accuracy using MedBullets-derived questions and then addresses multi-turn clinical interactions through 5.3k generated scenarios with multi-dimensional rubrics.

media Hugging Face Forums · 2d ago · 18 views

MB-Bidram shares experimental WideNDepth architecture separating reasoning from knowledge

MB-Bidram has released an experimental neural architecture called WideNDepth (WND) designed to separate reasoning capabilities from knowledge storage. The model consists of a "Wide part" that functions as memory and a "Depth part" that acts as the reasoner.

media Hugging Face Forums · 3d ago · 18 views

Proposal: Progressive knowledge transfer from Qwen 9B to 27B for a fixed 4B model

A user on the Hugging Face forums proposes an experiment in progressive knowledge transfer, where a fixed-size student model (Qwen 4B) learns sequentially from increasingly larger teacher models rather than directly from the largest available one.

arxiv arXiv cs.LG · 7d ago · 20 views

OpenAI releases Open-1B with fully auditable training run

OpenAI has released Open-1B, a language model trained under a regime where every operation during training is independently reproducible on heterogeneous commodity hardware with bitwise certainty. This approach addresses the reproducibility issues inherent in open-source models by imposing a definite order on sources of nondeterminism, such as GPU kernel reductions and data batch ordering.