dstack
dstack copied to clipboard
An open-source container orchestration engine for running AI workloads in any cloud or data center. https://discord.gg/u8SmfwPpMd
For backward compatibility, let's add `DSTACK_RUN_NAME`. Once it's done, document all `DSTACK_*` variables.
Currently, dstack server provides API to create/update backends. This is supposed to be used by UIs/CLIs that provide alternative to `~/.dstack/server/config.yml`. But implementing an interface on top of this API...
This issue outlines the major items planned for Q2 2024. Note that it doesn't include bug fixes, except for major issues. > [!NOTE] > **Bold** means priority. ### Core features...
Currently, dstack AWS gateways are instances with public IP addresses. Users may want to create an internal-facing gateway so that the gateway is only accessible via internal IP inside the...
### Steps to reproduce If I run a service with model mapping without specifying chat_template, dstack goes to huggingface to get the tokenizer info. But if a model is restricted...
### Steps to reproduce - Make intentional error in the `docker args` script. For example, add the command `exit 1` or `kurl "https://dstack.ai"` to the script - Run `dstack run`...
### Problem Currently, dstack lacks built-in functionality that would allow users to persist data between runs. Cloud providers usually provide data persistence via network volumes. The proposal is to introduce...
### Problem With AWS, [Capacity Blocks for ML](https://docs.aws.amazon.com/AWSEC2/latest/UserGuide/ec2-capacity-blocks.html) is the main way to access H100 today. ### Solution _No response_ ### Workaround _No response_ ### Would you like to help...
1. `spot_policy: spot|on-demand|auto` -> `spot: true|false|auto` 2. `retry_policy: {retry: true|false, duration: int|str}` -> `retry_duration: int|str|off` 3. `creation_policy: create|reuse|reuse-or-create` -> `reuse: false|true|auto` 4. `termination_policy: dont-destroy|destroy-after-idle` & `termination_idle_time: int|str` -> `idle_duration: int|str|off`...
### Steps to reproduce Step 1: Specify fix disk in .dstack.yml eg: ``` type: dev-environment # Use either `python` or `image` to configure environment python: "3.11" # image: ghcr.io/huggingface/text-generation-inference:latest ide:...