3 papers
cs.AI2026
Shared-Prefix KV Reuse Across Standard LoRA Adapters: Quality and Serving Tradeoffs
Dushyant Rajput
A common small-model deployment runs one shared backbone with several LoRA specialists that answer over the same context. Serving them naively re-prefills that shared context once…
cs.AI2026
Depth and Scale in the Sub-150M Regime: JugnuLM-53M vs JugnuLM-110M
Dushyant Rajput, Nirdesh Chauhan, Siddharth Kosaraju
We scale our conventional sub-150M pretraining recipe from 53.5M to 109.7M parameters, holding the method fixed (Qwen3-style decoder with grouped-query attention, RoPE, SwiGLU, RMS…
cs.AI2026
Cheap Verifiers, Large Blind Spots: Measuring the Reliability Cost of Cost-Saving Cascades
Dushyant Rajput, Nirdesh Chauhan, Siddharth Kosaraju
Inference cascades cut cost by answering most queries with a cheap model and escalating a hard tail to a frontier model that acts as verifier. A natural extension closes the loop:…