graphcore-research/llm-inference-research

Issues

• It seems that simply remove the local windows leads to better performance ?

  #3 opened 2 months ago by sxbkdby8
  2

• End-to-end Inference Time Benchmarking

  #2 opened 4 months ago by prajwal1210
  1

• Plans to release Triton SparQ?

  #1 opened 4 months ago by liyucheng09
  1