Cerebras Hits 2,100 Tokens/Second on Llama 3.1 70B
Cerebras said its wafer-scale platform tripled to 2,100 tokens per second on Llama 3.1 70B and ran 16 times faster than the quickest GPU setup.
Cerebras said its wafer-scale platform tripled to 2,100 tokens per second on Llama 3.1 70B and ran 16 times faster than the quickest GPU setup.