5 papers
Orchestration for Domain-specific Edge-Cloud Language Models
Prasoon Patidar, Alex Crown, Kevin Hsieh +4
The remarkable performance of Large Language Models (LLMs) has inspired many applications, which often necessitate edge-cloud collaboration due to connectivity, privacy, and cost c…
Direct Reasoning Optimization: Token-Level Reasoning Reflectivity Meets Rubric Gates for Unverifiable Tasks
Yifei Xu, Tusher Chakraborty, Srinagesh Sharma +6
Reinforcement learning (RL) training of large language models (LLMs) on unverifiable tasks is challenging even when a reasonable-quality reference answer is available. We propose a…
Diagnosing Capability Gaps in Fine-Tuning Data
Saeid Asgari Taghanaki, Rakshanda Agarwal, Bruce Sun +10
Fine-tuning large language models (LLMs) for domain-specific tasks requires training datasets that comprehensively cover the target capabilities a practitioner needs. Yet identifyi…
SibylSense: Adaptive Rubric Learning via Memory Tuning and Adversarial Probing
Yifei Xu, Guilherme Potje, Shivam Shandilya +9
Designing aligned and robust rewards for open-ended generation remains a key barrier to RL post-training. Rubrics provide structured, interpretable supervision, but scaling rubric…
RLTHF: Targeted Human Feedback for LLM Alignment
Yifei Xu, Tusher Chakraborty, Emre Kıcıman +11
Fine-tuning large language models (LLMs) to align with user preferences is challenging due to the high cost of quality human annotations in Reinforcement Learning from Human Feedba…