Code

[LLM] llm.c training for GPT 2#3611

Code

Merged

Michaelvll merged 32 commits intomasterskypilot-org/skypilot:masterfrom gpt-2skypilot-org/skypilot:gpt-2Copy head branch name to clipboardMay 31, 2024

+335-2

Commits

Commits on May 28, 2024

• add gpt-2 example

  Michaelvll

  committed
  374f0f2

Commits on May 29, 2024

• Use ubuntu for GCP

  Michaelvll

  committed
  79323f7

• fix ncl

  Michaelvll

  committed
  03623ee

• Fix GPT-2

  Michaelvll

  committed
  3636ea6

• add train and data

  Michaelvll

  committed
  1694ecd

• use 8 gpus

  Michaelvll

  committed
  2c80dcb

• revert gcp change

  Michaelvll

  committed
  1bef798

• update readme

  Michaelvll

  committed
  9282873

• Add GCP image

  Michaelvll

  committed
  0ee942c

• make file_mounts more general

  Michaelvll

  committed
  5af0d93

• avoid any_of

  Michaelvll

  committed
  71bcdd0

Commits on May 30, 2024

• change back to use ubuntu image with wait for GPU

  Michaelvll

  committed
  488347f

• Merge branch 'gpt-2' of https://github.com/skypilot-org/skypilot into gpt-2

  Michaelvll

  committed
  8ec06a8

• wait cuda installation

  Michaelvll

  committed
  2e5bacf

• Add retry for file mount and use env for bucket name

  Michaelvll

  committed
  c070da0

• revert retries

  Michaelvll

  committed
  87d2a3c

• update the image

  Michaelvll

  committed
  d6e9554

• Merge branch 'master' of https://github.com/skypilot-org/skypilot into gpt-2

  Michaelvll

  committed
  0c2d799

• change to docker for better dependency

  Michaelvll

  committed
  ef26ecd

• revert changes in gcp template

  Michaelvll

  committed
  2b0a085

• avoid using docker on lambda

  Michaelvll

  committed
  aa8ecfe

• Add single GPU

  Michaelvll

  committed
  265e43c

• Elaborate readme

  Michaelvll

  committed
  598dca5

• Update llm/gpt-2/README.md

  Show description for 3056c2c
  

  Michaelvll

  and

  romilbhardwaj

  authored
  3056c2c

• fix

  Michaelvll

  committed
  815d23c

• Merge branch 'gpt-2' of https://github.com/skypilot-org/skypilot into gpt-2

  Michaelvll

  committed
  faf63d8

Commits on May 31, 2024

• address comments

  Michaelvll

  committed
  4c44935

• Fix data fetching

  Michaelvll

  committed
  3b7312e

• Add visualization

  Michaelvll

  committed
  b6566d7

• update

  Michaelvll

  committed
  bea72d5

• reduce cpu cost

  Michaelvll

  committed
  8887435

• update loss curve

  Michaelvll

  committed
  7609990