You signed in with another tab or window. Reload to refresh your session.You signed out in another tab or window. Reload to refresh your session.You switched accounts on another tab or window. Reload to refresh your session.Dismiss alert
Docker image bundling[Ollama](https://ollama.com/)with the [Sprocket](https://github.com/stjude-rust-labs/sprocket) WDL validator, designed for benchmarking LLM-generated WDL scripts.
3
+
This directory contains Docker images for[Ollama](https://ollama.com/), an LLM inference server, bundled with the [Sprocket](https://github.com/stjude-rust-labs/sprocket) WDL validator and the [Python ollama SDK](https://pypi.org/project/ollama/). Designed for benchmarking LLM-generated WDL scripts.
4
4
5
5
## Available Versions
6
6
7
-
| Tag | Ollama Version | Sprocket Version | Python ollama SDK |
These Docker images are built from `ollama/ollama:0.21.0` and include:
13
+
14
+
- Ollama v0.21.0: LLM inference server for running models locally
15
+
- Sprocket v0.23.0: WDL script validator (compiled from source via cargo)
16
+
- Python ollama SDK v0.6.1: Python client library for interacting with Ollama
17
+
- Python 3 (system version from base image)
18
+
19
+
Note: Sprocket is compiled from source during the Docker build using Rust's `cargo install`. The Rust toolchain is removed after compilation to minimize image size.
11
20
12
21
## Platform Availability
13
22
@@ -17,25 +26,48 @@ Sprocket is compiled from source during the Docker build, which causes arm64 bui
17
26
18
27
A GPU is not required to run this image, but is highly encouraged — CPU-only execution of LLMs is significantly slower.
19
28
29
+
## Citation
30
+
31
+
This image bundles two independent tools. If you use them in your research, please cite the original authors:
docker run --rm getwilds/ollama:latest ollama --version
69
+
docker run --rm getwilds/ollama:latest sprocket --version
70
+
39
71
# Start the container with GPU access
40
72
docker run --rm --gpus all -it getwilds/ollama:latest
41
73
@@ -57,20 +89,25 @@ with open('/tmp/output.wdl', 'w') as f:
57
89
sprocket lint /tmp/output.wdl
58
90
```
59
91
60
-
## Installed Components
92
+
## Dockerfile Structure
61
93
62
-
- Ollama: v0.21.0 (LLM inference server)
63
-
- Sprocket: v0.23.0 (WDL script validator, compiled from source via cargo)
64
-
- Python ollama SDK: v0.6.1
65
-
- Python 3 (system version from base image)
94
+
The Dockerfile follows these main steps:
66
95
67
-
Note: Sprocket is compiled from source during the Docker build using Rust's `cargo install`. The Rust toolchain is removed after compilation to minimize image size.
96
+
1. Uses `ollama/ollama:0.21.0` as the base image
97
+
2. Adds metadata labels for documentation and attribution
98
+
3. Installs system dependencies with pinned versions (Python, curl, build tools)
99
+
4. Installs the Python ollama SDK via pip
100
+
5. Installs Rust via rustup, compiles Sprocket from source, then removes the Rust toolchain
101
+
6. Runs smoke tests to verify all tools are installed correctly
102
+
103
+
## Security Scanning and CVEs
104
+
105
+
These images are regularly scanned for vulnerabilities using Docker Scout. However, due to the nature of bioinformatics software and their dependencies, some Docker images may contain components with known vulnerabilities (CVEs).
68
106
69
-
## Security
107
+
**Use at your own risk**: While we strive to minimize security issues, these images are primarily designed for research and analytical workflows in controlled environments.
70
108
71
-
Vulnerability reports are available in this directory as `CVEs_*.md` files.
72
-
Images are scanned monthly and on each build.
109
+
For the latest security information about this image, please check the `CVEs_*.md` files in [this directory](https://github.com/getwilds/wilds-docker-library/blob/main/ollama), which are automatically updated through our GitHub Actions workflow. If a particular vulnerability is of concern, please file an [issue](https://github.com/getwilds/wilds-docker-library/issues) in the GitHub repo citing which CVE you would like to be addressed.
73
110
74
-
## Contributing
111
+
## Source Repository
75
112
76
-
See the [CONTRIBUTING.md](../.github/CONTRIBUTING.md) for guidelines.
113
+
These Dockerfiles are maintained in the [WILDS Docker Library](https://github.com/getwilds/wilds-docker-library) repository.
0 commit comments