Скачать или смотреть Scandals in AI: Objaverse, Llama, Alpaca, and Dolly

My take on Objaverse, Llama, Alpaca, and Dolly. Not a lot of respect for copyright or contract terms. #largelanguagemodels #datascience #machinelearning #objaverse #alleninstitute #openai #chatgpt #llama #alpaca #meta #databricks

References:
Objaverse:   / 1638608472525897728  
Sketchfab on Objaverse:   / 1639311132253995023  

Llama:   / large-language-model-llama-meta-ai   & https://huggingface.co/decapoda-resea...

Alpaca: https://github.com/tatsu-lab/stanford... & https://crfm.stanford.edu/2023/03/13/...

Dolly: https://www.databricks.com/blog/2023/... https://github.com/databrickslabs/dolly

Databricks could have spent 50k and built a new dataset that would have contributed to the community. Instead, they used a dataset that violated the terms of the contract of their competitor, OpenAI. The model Databricks touted offers minimal value to the community (fine-tuning an existing dataset on an existing model). NLP Cloud previously wrote and shared about their model that used the Alpaca dataset on GPT-J well before Databricks published Dolly.

UPDATE: On April 12th, Databricks shared a new version of Dolly using a new dataset they created. The new model is now available for commercial use.

NLPCloud (which did this before Databricks): https://huggingface.co/nlpcloud/instr...
https://nlpcloud.com/instruct-version...

━━━━━━━━━━━━━━━━━━━━━━━━━
★ Rajistics Social Media »
● Link Tree: https://linktr.ee/rajistics
● LinkedIn:   / rajistics  
━━━━━━━━━━━━━━━━━━━━━━━━━