Accompaniment to this video where I test Llama 3 7b against a bunch of models. Here's a bunch of relatively tough questions I use to grade models and see how they perform!
Each folder in prompts has a bunch of task prompts and at least one source to use. Copy them together and you're done!