Eri is the open proof of process: pretrained from scratch on Navdyut’s scanned Assamese corpus, not adapted from a multilingual checkpoint that barely saw the script. Weights are on Hugging Face for research use.
The curve below is validation perplexity on a held-out Assamese literary and gazette mix during the final pretraining run.
Deep model documentation stays on the Models page. This post is the release record.