meta-llama/Llama-3.1-405B-Instruct-FP8 Benchmark | LLM Performance Benchmarks