issues
search
trapoom555
/
Language-Model-STS-CFT
Improving Text Embedding of Language Models Using Contrastive Fine-tuning
60
stars
7
forks
source link
STS Evaluation
#7
Closed
trapoom555
closed
6 months ago
trapoom555
commented
6 months ago
STS Benchmark
(validation set) can be used for evaluation during training
This benchmark can imply how well the model create an embedding vector given a sentence
Score will be calculated using Spearman Correlation