DOLLY 2 LLM defined: New Open-Supply LLM w/ code

New LLM by Databricks: Dolly 2.0 is a brand new 12B parameter language mannequin (LLM) primarily based on the EleutherAI Pythia mannequin household and instruction fine-tuned solely on a brand new, high-quality human generated instruction following dataset, crowdsourced amongst Databricks staff. Instruction fine-tuning with a open-source knowledge set, created by @Databricks

Plus PyTorch code implementation of DOLLY 2 on a free COLAB Pocket book.

“We (Databricks) are open-sourcing the whole lot of Dolly 2.0, together with the coaching code, the dataset, and the mannequin weights, all appropriate for business use. Which means any group can create, personal, and customise highly effective LLMs that may discuss to folks, with out paying for API entry or sharing knowledge with third events.”– Thanks, Databricks!

All rights with corresponding authors:

https://www.databricks.com/blog/2023/04/12/dolly-first-open-commercially-viable-instruction-tuned-llm

https://huggingface.co/databricks/dolly-v2-12b

https://github.com/EleutherAI/pythia

00:00 DOLLY 2.0 (open-source)02:39 15K Instruction dataset by DB (open-source)07:44 PYTHIA LLM08:37 CODE (PyTorch) COLAB

#ai#dolly2#databricks#datascience#opensource#naturallanguageprocessing

The post DOLLY 2 LLM defined: New Open-Supply LLM w/ code appeared first on AIPressRoom.