Uploaded image for project: 'Red Hat OpenShift AI Engineering'
  1. Red Hat OpenShift AI Engineering
  2. RHOAIENG-6839

Update TGIS Image for 2.8.3 to include mlp and attn bias option for flash and paged llama models

XMLWordPrintable

    • 2
    • False
    • Hide

      None

      Show
      None
    • False
    • No
    • No
    • Model Serving Sprint Q2-3
    • Testable

      PR from IBM upstream that needs to be included in the new 2.8.3 TGIS image: https://github.com/IBM/text-generation-inference/pull/85

      The ticket is to:

      1. Cherry-pick the ibm/tgis PR to rhds/tgis:rhoai-2.8
      2. Ensure the image is built and pushed to modh/tgis
      3. Update the rhds/odh-dashboard:rhoai-2.8 branch with new image

      Note: The inclusion of this particular PR in 2.10 will be done as part of a regular image-syncing activity for all runtimes for 2.10 release.

            selbi@redhat.com Selbi Nuryyeva
            selbi@redhat.com Selbi Nuryyeva
            Berto D'Attoma, Lucas Fernandez Aragon, Sean Pryor, Tarun Kumar
            RHOAI Model Serving Runtimes
            Votes:
            0 Vote for this issue
            Watchers:
            5 Start watching this issue

              Created:
              Updated: