Skip to content
GitLab
Projects Groups Snippets
  • /
  • Help
    • Help
    • Support
    • Community forum
    • Submit feedback
    • Contribute to GitLab
  • Sign in / Register
  • M metaseq
  • Project information
    • Project information
    • Activity
    • Labels
    • Members
  • Repository
    • Repository
    • Files
    • Commits
    • Branches
    • Tags
    • Contributors
    • Graph
    • Compare
  • Issues 95
    • Issues 95
    • List
    • Boards
    • Service Desk
    • Milestones
  • Merge requests 41
    • Merge requests 41
  • CI/CD
    • CI/CD
    • Pipelines
    • Jobs
    • Schedules
  • Deployments
    • Deployments
    • Environments
    • Releases
  • Packages and registries
    • Packages and registries
    • Package Registry
    • Infrastructure Registry
  • Monitor
    • Monitor
    • Incidents
  • Analytics
    • Analytics
    • Value stream
    • CI/CD
    • Repository
  • Wiki
    • Wiki
  • Snippets
    • Snippets
  • Activity
  • Graph
  • Create a new issue
  • Jobs
  • Commits
  • Issue Boards
Collapse sidebar
  • Administrator
  • metaseq
  • Merge requests
  • !226

[api] Fix bad batching causing wrong number of tokens returned or cutoff.

  • Review changes

  • Download
  • Email patches
  • Plain diff
Merged Administrator requested to merge nooverflow into main Jul 15, 2022
  • Overview 4
  • Commits 2
  • Pipelines 0
  • Changes 2

Created by: stephenroller

Patch Description There are some adversarial setups where a user could produce a long prompt (2040 prompt, 8 gen) and another user could produce a very short prompt with long generation (8 prompt, 2040 gen). These two might get batched together because individually they're below 2048. However, once padded, it becomes an input of 4080. This can cause one of the two outputs to be returned incorrect.y

Testing steps Manual testing

Assignee
Assign to
Reviewers
Request review from
Time tracking
Source branch: nooverflow