Bert Score for Contextual Similarity for RAG Evaluation

[@manishsharma2211]

Quick Correction : ROUGE is used for summarization and BLEU is used for translation primarily [ time stamp : 4:04 ]

[@AIAnytime]

My bad in rush. Thanks

[@SonGoku-pc7jl]

thanks for all! :) tomorrow i will watch your others videos of evaluate llm and rag :)

[@AIAnytime]

Rock on!

[@gotitgotya]

great work man....thank you so much for uploading such informative videos❤❤

[@VenkatesanVenkat-fd4hg]

Thanks for valuable video as usual. Waiting for multimodal and unstructured files/applying RAG

[@soumilyade1057]

The library that you have used mentions use of custom models in point 3 of ReadMe. But, there's no parameter by the name "model" or "num_layers". I was wondering if you have figured out what's going on there.

[@ShreyaSingh-wp9yk]

thanks for uploading the video. one quick question how it is different from rogue, bleu and meteor, as they are also recall and precision-based.can we use rogue, meteor and bertscore if I am evaluating chatbot and why. Please excuse if this ques sounds naive, I am very much new to this

[@ArpitBhavsar-z2c]

Though, in the second example, it gave 85% similarity
Weird right?

[@CibeSridharanK]

exactly thats why his reaction is strange

[@encianhoratiu5301]

yeah, not so good to evalute a llm generation... I don't blame him perahps there isn't a better way.

[@livelaughmotivate94]

For Text Generation RAG, Bert Score won't work?