Author

Praveen Oosa

h-index 0 View on Semantic Scholar

We have 1 of 1 papers

We haven’t gathered this author’s papers yet. Follow them and we’ll fetch their work.

Not the right person? Other researchers publish under this name.

Preprint Jul 2026

Wiola 13M, a Gated Spiral Attention Architecture for Parameter Efficient Small Language Models

Wiola is presented, a decoder only language model whose novelty is concentrated in three drop in components of every layer, and it is proved that the gated attention admits an exact and numerically verified equivalence between full sequence training and cached autoregressive decoding, so that no approximation is introd...

A. Chowdhury, Praveen Oosa, V. Reddy · 0 citations