uzlm/alloma-8B-Instruct
TEXT GENERATIONConcurrency Cost:1Model Size:8BQuant:FP8Ctx Length:32kPublished:Sep 8, 2025License:llama3.1Architecture:Transformer0.0K Cold

uzlm/alloma-8B-Instruct is an 8-billion parameter Llama-series instruction-tuned model developed by Examy.me and Teamwork.uz, optimized for the Uzbek language. It features a customized tokenizer that processes Uzbek text twice as fast as original Llama models, enabling longer effective context lengths. Continually pretrained on a bilingual dataset (67% English, 33% Uzbek), this model excels in Uzbek translation and sentiment analysis tasks.

Loading preview...