norcalnatvB to Hardware@hardware.watchEnglish · 2 years agoNvidia, Intel claim new LLM training speed records in new MLPerf 3.1 benchmarkventurebeat.comexternal-linkmessage-square8linkfedilinkarrow-up11arrow-down10
arrow-up11arrow-down1external-linkNvidia, Intel claim new LLM training speed records in new MLPerf 3.1 benchmarkventurebeat.comnorcalnatvB to Hardware@hardware.watchEnglish · 2 years agomessage-square8linkfedilink
minus-squarenorcalnatvOPBlinkfedilinkEnglisharrow-up1·2 years agoAmong many new records and milestones, one in generative AI stands out: NVIDIA Eos — an AI supercomputer powered by a whopping 10,752 NVIDIA H100 Tensor Core GPUs and NVIDIA Quantum-2 InfiniBand networking — completed a training benchmark based on a GPT-3 model with 175 billion parameters trained on one billion tokens in just 3.9 minutes.
minus-squareFlowerstar1BlinkfedilinkEnglisharrow-up1·2 years agoWow I wonder how long that would have taken in Turing/Volta era hardware.
minus-squareiDontSeedMyTorrentsBlinkfedilinkEnglisharrow-up1·2 years agoWhat does it mean to train on such-and-such number of tokens?
Among many new records and milestones, one in generative AI stands out: NVIDIA Eos — an AI supercomputer powered by a whopping 10,752 NVIDIA H100 Tensor Core GPUs and NVIDIA Quantum-2 InfiniBand networking — completed a training benchmark based on a GPT-3 model with 175 billion parameters trained on one billion tokens in just 3.9 minutes.
Wow I wonder how long that would have taken in Turing/Volta era hardware.
What does it mean to train on such-and-such number of tokens?