3 papers
cs.CL2026
Label Smoothing in Disguise? Gated Knowledge Distillation for Low-Resource Bangla Generation
Dipto Sumit, Zobeida Zaman Chowdhury, Syeda Fabiayee Jahan +1
Gated knowledge distillation (KD) weights teacher signals per token or example, aiming to trust an imperfect teacher only where it seems reliable. We test two gates: EWAD (entropy-…
cs.AI2026
Temperature, Not Knowledge: What Recursive Self-Training and Its Remedies Do to a Language Model's Confidence
Dipto Sumit, Syeda Fabiayee Jahan, Farig Sadeque
Recursive self-training can collapse language models, and many proposed remedies target overconfidence, yet their effects on confidence are rarely measured. We study a five-round s…
cs.AI2026
Reading Between the Chapters: A Graph-Free RAG Rebuilt With AI and Human Feedback
Dipto Sumit, Syeda Fabiayee Jahan, Farig Sadeque
Many questions about a book can only be answered by connecting passages chapters apart, such as a cough heard three times across a horse's life. We introduce CrossChapterQA, 1,853…