Llama2 13B finetuned to process longer (32K) contexts using interpolation and then further instruct finetuned with ShareGPT and Orca-Chat.
This is next version of the model trained and evaluated as part of the experiments described in the repo
http://github.com/abacusai/Long-Context. This version
was trained with a scaling factor of 8 and shows better reasoning and math abilites on State of the Art benchmarks
Since the model is instruct finetuned it can also be directly used for various prompted tasks. We have tested
it on open book question answering using the long context to supply search results.
The model has not been evaluated for safety and is only intended for research and experiments.
