thank you for this video, i'm working in internship on NLP retrieval model, i used GPL, i use the company data. i was wondering if MRR@10 equal 0.75 is acceptable or not. i did not yet find some MRR scores on the internet to describe GPL models
Great video! Since I am also looking at the SetFit notebook by Huggingface, I got a training related question. It appears to me that the Huggingface idea of a few shot training, is training on examples like text - label. For me it means they train with the classification head with the model body being frozen (bcs this is a siamese model so you need 2 text inputs to train it). If I want to train the body, i need examples as Nils is presenting. Do I miss anything here?
Two amazing people in one place, great talk Nils! :) Thanks for hosting Jay!
thank you for this video, i'm working in internship on NLP retrieval model, i used GPL, i use the company data. i was wondering if MRR@10 equal 0.75 is acceptable or not. i did not yet find some MRR scores on the internet to describe GPL models
Great video! Since I am also looking at the SetFit notebook by Huggingface, I got a training related question. It appears to me that the Huggingface idea of a few shot training, is training on examples like text - label. For me it means they train with the classification head with the model body being frozen (bcs this is a siamese model so you need 2 text inputs to train it). If I want to train the body, i need examples as Nils is presenting. Do I miss anything here?
Nils..!!!