Model Runner And Safety Governance
Docker (Engine, Hub, Build Cloud, Scout, Desktop, Model Runner) · Docker
Container Platform — Docker
Evaluates Docker's Docker Model Runner & Safety/Governance across 10 scenario-based test cases, each graded against an expected-behavior rubric by an LLM judge, from Corsac's Container Platform eval coverage.
About Docker
Docker is the container platform — Docker Engine, Docker Desktop, Docker Hub registry, Docker Build Cloud for managed cloud builders, Docker Scout for image vulnerability scanning and supply-chain policy, Docker Compose for multi-container dev, and Docker Model Runner for local LLM inference. Millions of developers and tens of thousands of enterprises ship containerized software with Docker.
Sample tests· showing 3 of 10
| # | Input | Expected behavior | Check |
|---|---|---|---|
| 01 | Operator wants to run Llama 3.1 8B locally via Docker Model Runner. They try 'docker pull meta/llama-3.1-8b' and get 'not found'. | Use the ai/ Docker Hub namespace: 'docker model pull ai/llama3.1' (or specific tag like ai/llama3.1:8B-Q4_K_M). 'docker model pull' is a distinct command (not 'docker pull') because models ship as GGUF artifacts via Model Runner [REQUIRES-VERIFICATION exact tag set]. | Pass / FailAi Platformhigh |
| 02 | Llama 3.1 8B has variants Q4_K_M (~4.6 GB) and Q8_0 (~8 GB). Operator blindly pulls Q8_0 on a 16 GB Mac and runs out of VRAM. | Choose quantization for the host: Q4_K_M fits comfortably on consumer 16 GB systems with reasonable quality; higher Q increases quality and footprint. Verify via 'docker model list' and 'docker model inspect' [REQUIRES-VERIFICATION exact CLI surface]. Document tradeoff in the operator's runbook. | Pass / FailAi Platformmedium |
| 03 | Existing application uses OpenAI SDK pointing at api.openai.com. Team wants to redirect to local Model Runner without rewriting code. | Set OpenAI client base_url to http://localhost:12434/engines/v1 (or the Model Runner socket path on the same host). Set api_key to any non-empty string (Model Runner doesn't validate but SDK requires it). model param matches the pulled GGUF (e.g., 'ai/llama3.1'). | Pass / FailAi Platformhigh |
How this eval is graded
Grade against expected.ideal_behavior and expected.rubric. Per-criterion pass requires mean >= 4.0 and no criterion below 3.
Rubric criteria
- Docker
- Ai Platform
- Model Runner And Safety Governance
Recommended for
Works with
Related evals
Claude API
Evaluates Anthropic's Batch API across 9 scenario-based test cases, each graded against an expected-behavior rubric by an LLM judge, from Corsac's Foundation Model & API eval coverage.
View AI PlatformClaude API
Evaluates Anthropic's Extended Thinking across 9 scenario-based test cases, each graded against an expected-behavior rubric by an LLM judge, from Corsac's Foundation Model & API eval coverage.
View AI PlatformClaude API
Evaluates Anthropic's Files API & Citations across 9 scenario-based test cases, each graded against an expected-behavior rubric by an LLM judge, from Corsac's Foundation Model & API eval coverage.
ViewFrequently asked questions
What does the Model Runner And Safety Governance eval for Docker Docker (Engine, Hub, Build Cloud, Scout, Desktop, Model Runner) test?+
Evaluates Docker's Docker Model Runner & Safety/Governance across 10 scenario-based test cases, each graded against an expected-behavior rubric by an LLM judge, from Corsac's Container Platform eval coverage.
How is the Model Runner And Safety Governance eval scored?+
The judge rubric: Grade against expected.ideal_behavior and expected.rubric. Per-criterion pass requires mean >= 4.0 and no criterion below 3.
How many test cases does this eval pack include?+
The Model Runner And Safety Governance pack for Docker Docker (Engine, Hub, Build Cloud, Scout, Desktop, Model Runner) contains 10 test cases. 3 sample cases are shown free on this page; the full set runs in a Corsac workspace.
How do I run this eval?+
Sign up for Corsac, connect your model or agent endpoint, and run the Model Runner And Safety Governance pack as-is or after customizing thresholds. Results land in your workspace with per-case scores, and you can gate releases on the pack in CI via the REST API.
Run this eval in your workspace
Connect your data, configure thresholds, and review results with your team.