← Azimuth Daily
Capturing data…
Capturing data…
FRI, 6 NOV · 97 ITEMS
"AssameseLM-E2B open-weight Assamese language model released" — arXiv cs.CL · AI & Technology
This is the release of AssameseLM-E2B, a 2-billion-parameter language model specifically for , an Indo-Aryan language spoken primarily in Assam, India. The model is designed to run on consumer-grade hardware (e.g., a single GPU) and is detailed in a preprint on arXiv by Pragyan Phukan. The release makes a state-of-the-art LLM for a relatively low-resource language publicly available, addressing the gap in NLP tools for Assamese.
The claim derives directly from the arXiv preprint itself, which is the primary source. The arXiv listing confirms the paper title and author, and the grounding independently verifies the publication date. As a primary source, the paper is authoritative about its own release and design decisions, though independent evaluation of the model's performance would require reading the full paper.
Genuinely new: the preprint was submitted on 2026-11-05 and is being reported on the same day.