Code

Support molmo in turbomind#2716

Code

Merged

lvhan028 merged 71 commits intoInternLM:mainInternLM/lmdeploy:mainfrom lvhan028:support-molmolvhan028/lmdeploy:support-molmoCopy head branch name to clipboardNov 14, 2024

+653-8

Commits

Commits on Sep 2, 2024

• initial moe support

  lzhangzz

  committed
  81458e9

Commits on Sep 29, 2024

• dynamic grouped gemm

  lzhangzz

  committed
  de31050

Commits on Sep 30, 2024

• benchmark

  lzhangzz

  committed
  9b71f34

Commits on Oct 8, 2024

• moe benchmark

  lzhangzz

  committed
  8538b2c

• moe sampling

  lzhangzz

  committed
  3b731e2

Commits on Oct 9, 2024

• split-k

  lzhangzz

  committed
  e3c2faa

Commits on Oct 10, 2024

• refactor tuning

  lzhangzz

  committed
  e99848b

• simplify

  lzhangzz

  committed
  2b67a53

• n-major weight

  lzhangzz

  committed
  5176288

• add num for MatrixLayout

  lzhangzz

  committed
  018c338

Commits on Oct 11, 2024

• packed rows

  lzhangzz

  committed
  fee20be

• packed cols

  lzhangzz

  committed
  9fcaabc

Commits on Oct 14, 2024

• dispatch for packed rows

  lzhangzz

  committed
  815c581

• w4a16 moe

  lzhangzz

  committed
  a4a81d9

Commits on Oct 15, 2024

• refactor model loading

  lzhangzz

  committed
  1db7fe1

Commits on Oct 16, 2024

• fix pytorch loader

  lzhangzz

  committed
  04fd8b4

Commits on Oct 17, 2024

• refactor

  lzhangzz

  committed
  b3ceb17

Commits on Oct 18, 2024

• dispatch w4a16 moe

  lzhangzz

  committed
  c8d4ed5

• fix loader

  lzhangzz

  committed
  3e355df

• add comment

  lzhangzz

  committed
  da97d5f

• Merge remote-tracking branch 'origin/main' into moe

  lzhangzz

  committed
  7a4d6fb

• fix msvc build

  lzhangzz

  committed
  4ca33a2

• fix msvc build

  lzhangzz

  committed
  75cf858

• fix msvc build

  lzhangzz

  committed
  ad26ada

• fix ut

  lzhangzz

  committed
  ce59d29

• fix ut

  lzhangzz

  committed
  36a4e4e

• fix p-lora

  lzhangzz

  committed
  bd089a9

Commits on Oct 21, 2024

• add all support arches

  lzhangzz

  committed
  ab49732

• minor

  lzhangzz

  committed
  7a6f6e4

• fix lint

  lzhangzz

  committed
  7c8148c

• fix lint

  lzhangzz

  committed
  918f1e7

• fix lint

  lzhangzz

  committed
  0642ebd

• fix ut

  lzhangzz

  committed
  c2bf33e

• bf16 support

  lzhangzz

  committed
  783b266

• minor

  lzhangzz

  committed
  07e15f9

Commits on Oct 22, 2024

• checkin molmo conversion

  lvhan028

  committed
  d12d64c

• add chat template

  lvhan028

  committed
  85edbf1

• refactor

  lzhangzz

  committed
  a8bfb12

• fix lint

  lzhangzz

  committed
  25a3320

• fix ut

  lzhangzz

  committed
  792c412

• Just for test: hardcode vocab_size

  lvhan028

  committed
  4f875fb

• minor

  lzhangzz

  committed
  b2c858b

• minor

  lzhangzz

  committed
  c98ca7f

• minor

  lzhangzz

  committed
  504093a

Commits on Oct 23, 2024

• fix inter_size config

  lzhangzz

  committed
  4769ef8

• load with non-standard filenames

  lzhangzz

  committed
  b1fa486

• fix loader

  lzhangzz

  committed
  678fbed

• merge pr#2621

  lvhan028

  committed
  427c4f2

• fix missing default param

  lzhangzz

  committed
  5e1c9d7

• defer the loading of misc weights for safetensors

  lzhangzz

  committed
  1c66685

• add embedding_size

  lvhan028

  committed
  c0635f7

• update

  lvhan028

  committed
  bd013e6

• Merge branch 'PR-2621' into support-molmo

  lvhan028

  committed
  045022b

• update

  lvhan028

  committed
  e29c72b

• tmp

  lvhan028

  committed
  9b68a68

Commits on Oct 24, 2024

• tmp

  lvhan028

  committed
  0a744fa

Commits on Oct 28, 2024

• merge main

  lvhan028

  committed
  55c32e1

Commits on Oct 29, 2024

• Merge branch 'main' into support-molmo

  lvhan028

  committed
  62260a2

Commits on Oct 30, 2024

• update molmo template

  lvhan028

  committed
  d548056

Commits on Nov 5, 2024

• vision embedding

  lvhan028

  committed
  e3c7e77

• fix

  lvhan028

  committed
  2e1aea5

• Merge branch 'main' into support-molmo

  lvhan028

  committed
  c155963

• update

  lvhan028

  committed
  8d8f8b9

• Merge branch 'main' into support-molmo

  lvhan028

  committed
  0bfcae6

• fix

  lvhan028

  committed
  fea887c

Commits on Nov 11, 2024

• Merge branch 'main' into support-molmo

  lvhan028

  committed
  47c00d2

• fix messages2prompt in templates

  lvhan028

  committed
  ce1f229

• fix order of out_messages

  lvhan028

  committed
  6fbb28b

Commits on Nov 12, 2024

• fix

  lvhan028

  committed
  262f548

Commits on Nov 13, 2024

• add user guide

  lvhan028

  committed
  7d200bf

• update is_supported

  lvhan028

  committed
  bf76d43