Blog: Add post on leveraging Katib for efficient RAG optimization. #161
Conversation
|
[APPROVALNOTIFIER] This PR is NOT APPROVED This pull-request has been approved by: The full list of commands accepted by this bot can be found here.
Needs approval from an approver in each of these files:
Approvers can indicate their approval by writing |
|
FYI @andreyvelich we can put this under GenAI page |
varshaprasad96
force-pushed
the
post/rag_katib
branch
from
00ac27d to
5f6d99f
Compare
|
I'm not able to replicate the build error locally, trying to dig into it. Looks like I maybe missing some configuration, if someone has seen this issue before, that'd be super helpful. |
|
@tarekabouzeid did you encounter any issues here? |
No, i haven't. |
There was a problem hiding this comment.
Thank you for this great blog post @varshaprasad96!
I left a few comments.
/assign @varodrig @kubeflow/wg-training-leads @Electronic-Waste @helenxie-bit @akgraner
Please help us with the review
| available in the [documentation][katib_api]. | ||
|
|
||
| ```yaml | ||
| apiVersion: "kubeflow.org/v1beta1" |
There was a problem hiding this comment.
@varshaprasad96 I am wondering whether SDK submission would be easier, since we don't need to explain how to build Docker image from your RAG pipeline?
https://www.kubeflow.org/docs/components/katib/getting-started/#getting-started-with-katib-python-sdk
There was a problem hiding this comment.
@andreyvelich we would still need to build the image and pass in it in: https://github.com/kubeflow/katib/blob/c18035e1041ca1b87ea7eb7c01cb81b5e2b922b3/sdk/python/v1beta1/kubeflow/katib/api/katib_client.py#L195
There was a problem hiding this comment.
Can you use PyTorch as base_image and install required packages as follows:
def tune(
base_image="docker.io/pytorch/pytorch:2.2.1-cuda12.1-cudnn8-runtime",
packages_to_install=["rag",....],
)
Signed-off-by: Varsha Prasad Narsing <[email protected]>
varshaprasad96
force-pushed
the
post/rag_katib
branch
from
5f6d99f to
7000943
Compare
Closes: #160