“We want to see the traces, which LLMs are being used, how many tokens are being generated. The speed, the inference benchmarks, that’s all really important for us and our clients, and W&B Weave has been great for answering all of that.”