Maybe Reflection-70B really works. But that doesn't change the fact that people massively shared something that didn't. The API was wrong. The HF model was wrong (see my own benchmarks ↓).
By
–
Maybe Reflection-70B really works. But that doesn't change the fact that people massively shared something that didn't. The API was wrong. The HF model was wrong (see my own benchmarks ↓).