CStanKonrad/long_llama
A 3B–7B parameter large language model designed to process extended context windows beyond standard LLM limitations.

Velocity · 7d
+1.4
★ / day
Trend
→steady
star history
LongLLaMA extends OpenLLaMA’s context length capabilities through Focused Transformer (FoT) fine-tuning, enabling the model to handle significantly longer input sequences than typical language models. The project provides model weights on HuggingFace and includes instruction-tuned variants (Instruct and Code versions) with Colab notebooks for easy experimentation.