neuron-explorer
Steering-first neuron explorer for Qwen3.5-4B: push every MLP neuron, keep every rollout, browse the results
Open-source systems engineering for Software-Defined Vehicle — SysML v2, MBSE, product-line engineering, digital continuity, and continuous compliance.
Developer toolsPython · C++GitHub @de4sdvOnlineOwner unconfirmed
Have you tried it? One tap is enough. No sign-in needed.
No updates detected yet. GitHub releases will appear here.
Owners of public repositories under a personal GitHub account can sign in to verify ownership and manage the introduction themselves. No administrator approval is needed.
Sign in, generate a verification file, and commit it to this repository’s default branch within 30 minutes. Files on other branches or forks do not count. Then check the file below to manage the project without administrator approval. Keep the file to retain your access.
If needed, you will go to sign-in. After signing in, return here and press your chosen verification button again.
Run /nomorevibe verify in your project folder to verify ownership and manage the introduction yourself.
Verify in one minuteAsk about other verification options
The maker's introduction consists of the line above. Read the repository README for more details. Repository README ↗
Loading the full README.
Read the full README ↗An open-source project for digitally engineered, configurable, continuously certifiable software-defined vehicle (SDV) product lines. DE4SDV applies model-based systems engineering with SysML v2, product-line engineering, digital continuity, simulation interoperability, and continuous compliance — so that SDV variability across application domains (such as ADAS, telematics, connectivity), middleware, hardware abstraction layer and adapter layer is modeled explicitly as configurable architectures, enabling systematic comparison of alternatives, transparent trade-off decisions, and lifecycle-wide assurance.
DE4SDV is a workstream within the INCOSE Automotive Working Group.
New here? Start with DE4SDV through one case study: the configurable SDV stack and product line, the emergency-braking capability that exercises it, how agents, a deterministic Method Check, and people share the engineering work, and small tasks you can contribute without knowing the whole project.
In the SDV future, OEM differentiation will extend beyond traditional choices such as color, trim packages, sensors, and assistance systems. A central differentiator will be certified freedom: the ability to customize vehicle features within clearly defined guardrails that preserve safety, security, and compliance. Equally important are openness and trust, including transparency over how vehicle and user data is handled, and meaningful user control over that data.
DE4SDV embraces ecosystem diversity rather than locking into a single stack: across subsystems, multiple open-source alternatives already exist, and the project models that variability.
The core artifact is the SysML v2 systems model under
textual-notation-of-model/. Explore
it in the live viewer:
python -m tools.sysml_html_viewer.serve --repo . --port 8787
# open http://127.0.0.1:8787/
See the model viewer how-to for every feature, or click Help / Elements in the viewer header (the guides are published on the viewer website too).
New here? The getting started guide walks you through: what DE4SDV works on, the three-system framing, repository layout, exploring the model, understanding the increment process, running the checks, and making a first contribution.
To read the model source itself, the model element guide explains the SysML v2 element kinds used across the model and why DE4SDV uses them — from the increment shell and views to the verification/evidence chain.
DE4SDV uses the ASELCM three-system framing to keep product, engineering, and ecosystem scopes distinct:
Inside the System 2 boundary:
Outside the System 2 boundary, but connected through interfaces:
Boundary rule: an element is part of System 2 only when DE4SDV governs its configuration, change control, and evidence or traceability obligations. Otherwise it is treated as an external system at the ecosystem boundary, integrated through explicit interface and assurance contracts. This boundary definition aligns the project with digital-twin reference-model practice by making the managed twin boundary explicit.
| Area | Purpose |
|---|---|
docs/ | Human-facing documentation: getting-started, guides, terminology, ADRs, runbooks, plans |
approach/ | Process set, framework, ontology, viewpoints |
methodologies/ | Method guidance and reusable engineering methods |
textual-notation-of-model/ | System model assets and examples |
model-based-product-line-engineering/ | Feature models, configurations, shared assets, product models |
implementation/ | Reference implementations with reproducible evidence (see implementation/README.md) |
deployment/ | Public service deployment assets and operations guidance |
compliance/ | Safety, security, UNECE, homologation-related evidence placeholders |
continuous-homologation/ | Continuous compliance and approval evidence workflow |
configuration-management/ | Baselines, change control, versioning, release evidence |
standards/ | Standards map and interpretation notes |
digital-continuity/ | OSLC, digital thread, traceability, lifecycle integration |
digital-twin/ | Digital twin concepts, parameters, and runtime alignment |
simulation/ | FMI/FMU/SSP simulation integration assets |
sysmlv2-api/ | SysML v2 API clients, examples, integration notes |
devsecops/ | CI/CD, security automation, SBOM, threat-model automation |
experiments/ | Spikes and exploration results (not maintained assets) |
tools/ | Scripts, utilities, and development tooling |
scripts/ | Repository checks, smoke tests, SysML validation wrapper |
tests/ | Test suite for repository scripts and tooling |
.github/agents/ |
The curated repository map lives in
docs/repository-tree.md. Run git ls-files when a
complete machine-readable inventory is needed.
This project welcomes contributions beyond code: documentation, examples, safety/security analysis, modeling patterns, feature models, test cases, issue triage, and standards mapping are all valuable.
CONTRIBUTINGAGENTS if you use an AI coding assistant.github/ISSUE_TEMPLATE/Use GitHub issues and pull requests for project work, questions, proposals,
and decisions that need a durable public record. DE4SDV also uses Mattermost
for lightweight coordination at https://chat.de4sdv.org (invite-only;
request an invite through the Mattermost invite request issue template).
See COMMUNICATION for the full communication policy.
Except where otherwise noted, original DE4SDV content is licensed under the
Apache License 2.0. This repository also contains third-party or
derived material under other licenses; in particular, the generated
COVESA_VSS.sysml
library is licensed under the Mozilla Public License 2.0. Per-file SPDX
identifiers and accompanying license files identify the license applicable to
such material. See NOTICE for the bundled-material summary.
See also docs/references/source-notes for detailed source attributions.
| Task-specific AI agent instructions |