Q&A capability #147
Unanswered
jalkestrup
asked this question in
Q&A
Replies: 0 comments
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
-
Hi AI4Finance - first of all thanks for the awesome work being put into FinGPT and all the documentation shared.
As I understand, most of the recent models, including the https://huggingface.co/FinGPT/fingpt-mt_llama2-7b,_lora is finetuned on the entire Fin dataset, including this Q&A task, https://huggingface.co/datasets/FinGPT/fingpt-fiqa_qa , that has fairly long string outputs as the objective.
How come the current model only seem to support 'Financial Sentiment Analysis', 'Financial Relation Extraction', 'Financial Headline Classification' & 'Financial Named Entity Recognition'? - and provide very brief yes/no answers to any prompt?
For example, if I feed one of the training examples from the QA dataset,
And the output is: "No"
Other similar prompts provide equally short yes/no outputs.
Thanks for helping clarify this,
Beta Was this translation helpful? Give feedback.
All reactions