Bias Detection and Analysis in Transformer-based Language Models using Semantic and Layer-wise Representations | IJSREM