I’m curious to know if I can generate specific text with GPT-3 for Nasdaq stocks. I plan to use transfer learning for the following:-
- Scrape last 10 years results for Nasdaq stocks.
- Feed this customised dataset and perform transfer learning on GPT-3
- Use the new trained model and generate stock specific news.
I’m unable to find any references on transfer learning with GPT-3
I look forward to your response, TIA
We currently don’t have a feature for transfer learning, but that’s an interesting idea!
Bro, I really need to get access to the RAW matrix values. I can fuse data domains using the Huggingface (https://huggingface.co/) however that would be lightyears more powerful if I could just use OpenAI API to encode and them decode with my own APIs.
On the end of the day it is just math and data. There has been lots of ongoing studies from colleagues into fusing data domains, it seems this API is missing a potential right there: giving access to the matrix among some other “low-level” stuff.