-
Notifications
You must be signed in to change notification settings - Fork 91
Conversation
Signed-off-by: Sihan Wang <[email protected]>
Signed-off-by: Sihan Wang <[email protected]>
Signed-off-by: Sihan Wang <[email protected]>
Signed-off-by: Sihan Wang <[email protected]>
Signed-off-by: Sihan Wang <[email protected]>
Co-authored-by: shrekris-anyscale <[email protected]> Signed-off-by: Sihan Wang <[email protected]>
Co-authored-by: shrekris-anyscale <[email protected]> Signed-off-by: Sihan Wang <[email protected]>
Co-authored-by: shrekris-anyscale <[email protected]> Signed-off-by: Sihan Wang <[email protected]>
Co-authored-by: shrekris-anyscale <[email protected]> Signed-off-by: Sihan Wang <[email protected]>
Co-authored-by: shrekris-anyscale <[email protected]> Signed-off-by: Sihan Wang <[email protected]>
Co-authored-by: shrekris-anyscale <[email protected]> Signed-off-by: Sihan Wang <[email protected]>
Co-authored-by: shrekris-anyscale <[email protected]> Signed-off-by: Sihan Wang <[email protected]>
Signed-off-by: Sihan Wang <[email protected]>
Signed-off-by: Sihan Wang <[email protected]>
Co-authored-by: shrekris-anyscale <[email protected]> Signed-off-by: Sihan Wang <[email protected]>
Signed-off-by: Sihan Wang <[email protected]>
… into 0.5.0_release2
There was a problem hiding this comment.
Choose a reason for hiding this comment
The reason will be displayed to describe this comment to others. Learn more.
Approving the README
and config files.
How do i install tensorrt-llm in the ray-llm docker? I've tried to install tensorrt-llm several time but always stuck in the errors related to MPI or mpi4py |
Hi @rifkybujana, the trtllm will be installed by default inside the 0.5.0 image. Currently you can't install by yourself, we are planning to relax this restriction in the later version of rayllm. |
Hi @sihanwang41, thanks for the reply. When will the 0.5.0 image be released? Also which version of TensorRT-LLM would be installed? |
it will be out this week, 0.6.1 version will be installed. |
Signed-off-by: Sihan Wang <[email protected]>
Signed-off-by: Sihan Wang <[email protected]>
There was a problem hiding this comment.
Choose a reason for hiding this comment
The reason will be displayed to describe this comment to others. Learn more.
One comment, otherwise looking good to me.
@@ -13,19 +11,17 @@ Jinja2 | |||
numexpr>=2.7.3 | |||
hf_transfer | |||
evaluate | |||
bitsandbytes | |||
vllm>=0.2.0 |
There was a problem hiding this comment.
Choose a reason for hiding this comment
The reason will be displayed to describe this comment to others. Learn more.
don't we need to pin this to <2.7?
<img width="1253" alt="Screen Shot 2023-06-05 at 4 35 41 PM" src="https://github.com/anyscale/aviary/assets/20109646/9e71db45-dd3b-4fb8-88f8-2ec28a78ae6e"> Follow up: * Autoscaler: The current image is missing some dependencies for KubeRay autoscaling. * Frontend: The frontend cannot be launched directly due to some dependency issues (e.g. `gradio`, `pymongo`, `boto3`...). --------- Co-authored-by: Antoni Baum <[email protected]>
Followup to #111 which adds a tutorial for GKE (the prior PR was for Amazon EKS). After all comments are addressed, before merging I will run through the tutorial again manually to make sure it works. - [ ] Final manual test "Fast follow" followups include: - [ ] Load test - [ ] Multiple models - [ ] Link to the tutorial from somewhere to make it discoverable. (Link to it from https://github.com/anyscale/aviary/blob/master/README.md I guess?) - [ ] Add a production guide using [RayService](https://ray-project.github.io/kuberay/guidance/rayservice/) More followups copied from #111: > Autoscaler: The current image is missing some dependencies for KubeRay autoscaling. > Frontend: The frontend cannot be launched directly due to some dependency issues (e.g. gradio, pymongo, boto3...). --------- Signed-off-by: Archit Kulkarni <[email protected]> Co-authored-by: Antoni Baum <[email protected]>
No description provided.