FlashAttention-2 Surpassed, 0.33 TB/s Linear O(n) Attention (Alphabase by SCS) (alphabase-bench.emergent.host) 1 points by GeometryKernel 5mo ago ↗ HN
[–] GeometryKernel 5mo ago ↗ Announcing Alphabase by Source Code SystemsURL: https://alphabase-bench.emergent.hostRESULT: FlashAttention-2 Surpassed, 0.33 TB/s Linear O(n) AttentionRun independent live benchmark test to 1 Trillion Tokens using Benchmark tool.Customize your own Live Benchmark Test, Download the Docker...orBuild and Deploy your own Fano-E8 LLM ...today.#AI #LLM #MachineLearning #DeepLearning #Transformers #ArtificialIntelligence #Tech #Startups #OpenSource #GPU #FlashAttention #AttentionMechanism #Inference #H100 #A100 #TBps #Throughput #Benchmark #MLops #MLPerf #Emergent #AIInfra #ServerlessGPU #RTX5090
1 comment
[ 1.5 ms ] story [ 14.2 ms ] threadURL: https://alphabase-bench.emergent.host
RESULT: FlashAttention-2 Surpassed, 0.33 TB/s Linear O(n) Attention
Run independent live benchmark test to 1 Trillion Tokens using Benchmark tool.
Customize your own Live Benchmark Test, Download the Docker...or
Build and Deploy your own Fano-E8 LLM ...today.
#AI #LLM #MachineLearning #DeepLearning #Transformers #ArtificialIntelligence #Tech #Startups #OpenSource #GPU #FlashAttention #AttentionMechanism #Inference #H100 #A100 #TBps #Throughput #Benchmark #MLops #MLPerf #Emergent #AIInfra #ServerlessGPU #RTX5090