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.
### Steps to reproduce 1. Create a task configuration with `env` setting as a list of strings: ``` env: - VAR-1=value - VAR:2=value ``` 2. Apply the configuration: `dstack apply...
Fine-tuning - [x] Alignment Handbook - [x] Axolotl - [ ] QLoRA - [ ] Pytorch Distributed Inference - [ ] Infinity - [ ] LoRAX - [ ] Ollama...
GitHub offers a very convenient syntax for commands: ```yaml jobs: bash-example: runs-on: ubuntu-latest steps: - name: Group of log lines run: | echo "::group::My title" echo "Inside group" echo "::endgroup::"...
1. For both `dstack ps` and `dstack fleet`, without `--verbose`, replace `RESOURCES` and `SPOT` with `GPU` (using the notation that the user specifies the GPU requirements, etc. `A10:24GB:3 (spot)`) 2....
Based on experience of using `dstack apply`, realized that creation of new fleets should always be explicit. By default, `dstack` should use the existing fleets. This makes it more predictable...
#3309 Close #3399
This issue outlines the major items planned for Q4 2025: High priority: - [x] #3124 - [ ] #2973 - [ ] #3126 - [ ] Sub-projects (allow share SSH...
**Steps To Test** Step1: Create `replica-groups-service.yml` ``` # replica-groups-service.yml type: service name: replica-groups-test python: 3.12 replica_groups: - name: replica-1 replicas: 0..2 scaling: metric: rps target: 2 commands: - echo "Group...
Implement the "UI" section from the parent issue design document. See the "Main concepts" and "API" sections for context. ## Implementation plan - [x] Events page - [ ] References...