NousResearch has announced the release of Genstruct 7B, an instruction-generation model that creates valid instructions from raw text corpuses. Led by @Euclaise_, the model can generate new, partially synthetic instruction finetuning datasets from various sources like Wikipedia or tweets. Other users are also experimenting with the model, such as vanstriendaniel who created cosmopedia_chat using Genstruct to generate user/assistant response pairs.
Introducing cosmopedia_chat (v0.0.1), my first experiment using the new @NousResearch Genstruct model. This dataset uses a subset of Cosmopedia, a synthetic textbook-quality dataset, and Genstruct to generate user/assistant response pairs. https://t.co/Wi9yOGUkAm
Another nice tool to make partially synthetic instruction finetuning datasets! https://t.co/N7o8fzDSVN
We're still cooking at Nous and today we're releasing a new model designed exclusively to create instructions from raw-text corpuses, Genstruct 7B. Led by @Euclaise_, this model is build to take in any raw passage from a text, i.e. a wikipedia entry or a tweet, or whatever you… https://t.co/xzTA7NdSML https://t.co/0MKMEYlWCH
We're still cooking at Nous and today we're releasing a new model designed exclusively to create instructions from raw-text corpuses, Genstruct 7B. Led by @Euclaise_, this model is build to take in any raw-text, i.e. a wikipedia entry or a tweet, or whatever you can come up… https://t.co/w04nq3xZam https://t.co/0MKMEYlWCH
Today we are announcing the release of a new model, Genstruct 7B. Genstruct 7B is an instruction-generation model, designed to create valid instructions given a raw text corpus. This enables the creation of new, partially synthetic instruction finetuning datasets from any… https://t.co/vFbipWpN2C