3 papers
cs.LG2026
NanoForecast v0.5: Competitive Time Series Forecasting Through Training Pipeline Optimization
Gautam Kishore
We present NanoForecast v0.5, a 6.5M-parameter forecaster that competes with models 31x its size (TimesFM, 200M parameters) after training pipeline fixes and no architecture change…
cs.AI2026
Safe Error Correction for Language Models: Frozen-Base Adjustment with Capability Preservation
Gautam Kishore
We study a practical question: can a small correction module fix errors in a frozen language model's outputs without degrading its base capabilities? We propose CRN v2, a lightweig…
cs.LG2026
pico-type: A 1.5M-Parameter Byte-Level Multi-Head Content Classifier
Gautam Kishore
We introduce pico-type, a byte-level multi-head content classifier with approximately 1.5 million parameters that simultaneously predicts seven content properties from raw UTF-8 by…