Aspect-based Sentiment Analysis with InstructABSA

[@rafasbueno04]

Amazing content, Robert! Thank you. Could I get a similar script in R as well? I am not sure if there are ABSA libraries and functions developed in R. Happy holidays!

[@RobWynn]

@@rafasbueno04 you could try to build your own for sure if you really want it. My guess would be the vast majority of these types of projects are going to be in Python

[@nobody84980]

Thank youuu it really helped!! I just have a question.. why did u continue fine tuning an already fine tuned model on the same data or did u fine tune it on more data?

[@RobWynn]

@@nobody84980 I further fine tuned it using a custom dataset from my restaurant. Doing that boosted accuracy on the test set. You definitely don't have to though, you can just use the model as-is. Thanks for the comment.

[@nobody84980]

@RobWynn perfect thank you! Appreciate ur reply I was thinking of fine tuning it on more general data reviews using gpt 4 or such since itll be able to generalize better .. thought to let u know itd be easier than manually labeling data even its on custom data but the tool is honestly very easy to use it wouldnt take that much time.. another problem is that when a review like " the food wasnt great but i like the vibe" .. vibe referes to the aspect of the restaurant or " it was too loud" or " couldnt wash my hands.. it was too smelly" which refers to bathroom.. im not sure if these aspects can be fine tuned tho an llm would be smart enough

[@roninababan1239]

NICE BRO,NICE SKILL