Refactor README.md #678
Refactor README.md #678
Conversation
|
Can we nest CPU, GPU and Autoprovisioning links underneath Clusters? I think it makes more sense since they refer to specific "cluster create" paths/options |
Added nesting. |
| XPK supports a variety of hardware accelerators. | ||
| | Accelerator | Type | Create Cluster | Create Workload | | ||
| |-------------|--------------------|---------------------------------------------------------------------------------------------------------------|-----------------------------------------------------------------------------------------------------------------| | ||
| | TPU | v4 | [docs](./docs/usage/clusters.md) | [docs](./docs/usage/workloads.md) | |
There was a problem hiding this comment.
One thing that I personally find a bit confusing, is that all TPU docs lead to the same page, so as a user if I click on the link next to the TPU v4 row, I expect to get instructions for TPU v4, but instead all are fixed to TPU v5 and the user may not notice that. Similarly for GPUs and H100. Not sure how we can fix that, I guess if there is some way to use url params it could work, simpler would probably be to go back to the list format that we had before and have general links for cluster and workload create.
There was a problem hiding this comment.
There is no docs specific to TPU, so that's why I link to generic ones. I think this is a good starting point and good organization method. We can later address by adding docs specific for TPU. IMO table is much cleaner than the list present before.
3 participants
Fixes / Features
This is an attempt of docs refactoring performed by Gemini.
Preview: https://github.com/AI-Hypercomputer/xpk/blob/scaliby/docs-refactor/README.md
Testing / Documentation
Docs update only.