Model A/B testing and similarity metrics

In this video, we build a model A/B testing web application with Arcee Conductor and Gradio. We also implement several similarity metrics (Jaccard, Cosine Similarity, Levenshtein, and Semantic Similarity), as well as a user feedback mechanism. Then, we run a few examples with SLMs and LLMs, showing how small models can generate answers that are extremely similar to those of much larger models.