6 papers · 1 filter
-: Circuit-Restricted Weight Arithmetic for Selective Refusal
Aditya Kasliwal, Pratinav Seth, Vinay Kumar Sankarapu
Modern deployments require LLMs to enforce safety policies at scale, yet many controls rely on inference-time interventions that add recurring compute cost and serving complexity.…
Faithfulness to Refusal: A Causal Audit of Neuron Selectors
Ananth Eswar, Pratinav Seth, Utsav Avaiya +1
Attribution scores increasingly identify which neuron rows of a language model matter for applications such as pruning, interpretability, and editing for safety, yet whether they i…
CuratorKIT : Data Curation and Synthetic Data Generation for LLM Post-Training
Soham Bhattacharjee, Karun Sharma, Vinay Kumar Sankarapu +1
Data curation is a critical part of post-training pipelines for large language models, yet existing tools often treat ingestion, deduplication, synthetic generation, and quality fi…
ALIGNBEAM : Inference-Time Alignment Transfer via Cross-Vocabulary Logit Mixing
Chirag Chawla, Pratinav Seth, Vinay Kumar Sankarapu
Domain fine-tuning degrades the safety of large language models: fine-tuned specialists readily comply with harmful prompts framed in domain language. Existing inference-time defen…
Provenance-Grounded Gating and Adaptive Recovery in Synthetic Post-Training Data Curation
Soham Bhattacharjee, Karun Sharma, Vinay Kumar Sankarapu +1
Synthetic post-training pipelines commonly filter generated samples with reward models or holistic LLM judges, yet two practices remain rarely examined together: whether the filter…
Beyond Uniform Credit: Causal Credit Assignment for Policy Optimization
Mykola Khandoga, Rui Yuan, Vinay Kumar Sankarapu
Policy gradient methods for language model reasoning, such as GRPO and DAPO, assign uniform credit to all generated tokens - the filler phrase "Let me think" receives the same grad…