Code

Fix LongRoPE KV Cache#1

Code

Closed

zelinms wants to merge 28 commits intoykim362:youki/debug-phimoeykim362/vllm:youki/debug-phimoefrom zelinms:fit-cluster-tests-fp8-longropezelinms/vllm-phi:fit-cluster-tests-fp8-longropeCopy head branch name to clipboard

+121-35

Commits

Commits on Jul 30, 2024

• tmp

  linzeqipku

  committed
  7a7ccb3

  Copy full SHA for 7a7ccb3

• tmp

  linzeqipku

  committed
  b07d6c5

  Copy full SHA for b07d6c5

• tmp

  linzeqipku

  committed
  693c6df

  Copy full SHA for 693c6df

• tmp

  linzeqipku

  committed
  477c792

  Copy full SHA for 477c792

Commits on Jul 31, 2024

• update metadata

  linzeqipku

  committed
  5dd7ffb

  Copy full SHA for 5dd7ffb

• update metadata

  linzeqipku

  committed
  ac7f93a

  Copy full SHA for ac7f93a

• update metadata

  linzeqipku

  committed
  51b736e

  Copy full SHA for 51b736e

• update device

  linzeqipku

  committed
  29e5c92

  Copy full SHA for 29e5c92

• update metadata

  linzeqipku

  committed
  8e2fde2

  Copy full SHA for 8e2fde2

• update metadata

  linzeqipku

  committed
  77d44d4

  Copy full SHA for 77d44d4

• max_seq_tokens_list is empty

  linzeqipku

  committed
  3842e4d

  Copy full SHA for 3842e4d

• mixtral to phimoe

  linzeqipku

  committed
  f75159c

  Copy full SHA for f75159c

• phimoe

  linzeqipku

  committed
  2c3177b

  Copy full SHA for 2c3177b

• cuda graph

  linzeqipku

  committed
  a5e5908

  Copy full SHA for a5e5908

• static flow

  linzeqipku

  committed
  ccd65b0

  Copy full SHA for ccd65b0

Commits on Aug 1, 2024

• tmp

  linzeqipku

  committed
  ff111d9

  Copy full SHA for ff111d9

• llama

  linzeqipku

  committed
  6e301c1

  Copy full SHA for 6e301c1

• tmp

  linzeqipku

  committed
  6d440cc

  Copy full SHA for 6d440cc

• tmp

  linzeqipku

  committed
  45a0828

  Copy full SHA for 45a0828

• capture

  linzeqipku

  committed
  08d4c41

  Copy full SHA for 08d4c41

• update log

  linzeqipku

  committed
  804acf7

  Copy full SHA for 804acf7

• tmp

  linzeqipku

  committed
  c67b7bd

  Copy full SHA for c67b7bd

• code style

  linzeqipku

  committed
  f24a5db

  Copy full SHA for f24a5db

• resolve conflict

  linzeqipku

  committed
  2c370dd

  Copy full SHA for 2c370dd

• tmp

  linzeqipku

  committed
  74e0e13

  Copy full SHA for 74e0e13

• tmp

  linzeqipku

  committed
  05b4b66

  Copy full SHA for 05b4b66

• tmp

  linzeqipku

  committed
  315ab72

  Copy full SHA for 315ab72

• tmp

  linzeqipku

  committed
  c17a344

  Copy full SHA for c17a344