dstack
dstack copied to clipboard
dstack is an open-source alternative to Kubernetes, designed to simplify development, training, and deployment of AI across any cloud or on-prem. It supports NVIDIA, AMD, and TPU.
OCI provides preemptible (spot) instances with a 50% discount for most CPU shapes and some GPU shapes. At the moment the discount is available for P100 and V100 GPUs, but...
Currently, the documentation refers to an environment variable `$DSTACK_ROLE` but doesn't mention that the user has to specify it (and how).
This is to be done in `0.19` as it can be backward incompatible.
### Context Many cloud providers bundle one or several (e.g. 16) local disks with some instance types. Local disks have these traits: - Physically attached to the host and hence...
Currently, we have to ways to configure backends: 1. Automatic - `dstack server` automatically detects credentials and creates `~/.dstack/server/config.yml` 2. Explicit - when the user manually create `~/.dstack/server/config.yml` **Problems:** 1....
#1323 added single-device TPU Pods support. Multi-device TPU Pods have not been supported because running multi-node tasks on them may require changes to dstack. Currently, dstack runs different jobs of...
**Steps to reproduce:** 1. Set up a host where the user requires a password for `sudo` 2. Invoke `dstack pool add-ssh` **Actual behavior:** The server logs: ``` {"message": "Try to...
Add documentation demonstrating how to use Task to convert Hugging Face models to TensorRT-LLM engines for high-throughput inference