Repository navigation
feat: standalone sixg-bench deployment with tiered optional packages - #36
Conversation
…CI matrix) Agent-Logs-Url: https://github.com/j143/6g/sessions/c122eeeb-d4f4-4093-b6c3-c37ed6997d6e Co-authored-by: j143 <53068787+j143@users.noreply.github.com>
Agent-Logs-Url: https://github.com/j143/6g/sessions/c122eeeb-d4f4-4093-b6c3-c37ed6997d6e Co-authored-by: j143 <53068787+j143@users.noreply.github.com>
|
error: consider using error: could not compile
WARNING: No output specified with docker-container driver. Build result will only remain in the build cache. To push result image into registry use --push or to load image into docker use --load
|
…or is_none_or Agent-Logs-Url: https://github.com/j143/6g/sessions/6d14baa4-4c38-4434-88ad-e7d531dea040 Co-authored-by: j143 <53068787+j143@users.noreply.github.com>
Fixed in commit
|
|
…ility Agent-Logs-Url: https://github.com/j143/6g/sessions/679cb7d3-aef0-4187-a767-673120053fef Co-authored-by: j143 <53068787+j143@users.noreply.github.com>
|
... Fixed in commit |
Turns the workspace into a clone-and-run bench. Core tier has zero native dependencies; ONNX inference, Python plots, and Level-2 baseline CSVs are opt-in.
Tier model
Validatechecks--baselines--plotting--onnxort-backed sentence-transformer (swaps simulation stub)New:
sixg-benchCLI binary (src/bin/bench.rs)Exit code is non-zero on any failure — usable in CI pipelines.
install.shcargo; installs viarustupif absentcargo build --releasewith appropriate--featuresbased on flags--onnx: guides model file placement (models/all-MiniLM-L6-v2.onnx), setsSIXG_ONNX_MODEL--baselines: setsSIXG_BASELINES=baselines/--plotting:pip install -r requirements-plot.txtCargo feature flags
Added to workspace
Cargo.toml:6g-aiand6g-semanticgain anonnxfeature flag. The existing deterministicOnnxModelsimulation remains the default; realort::Sessionis the intended swap behind#[cfg(feature = "onnx")].Docker
Dockerfile— multi-stage (rust:1.79-slim builder → debian:bookworm-slim runtime), core tierDockerfile.onnx— addslibonnxruntimelayerdocker-compose.yml— three services:bench,bench-baselines,bench-onnxPython plotting (
scripts/plot_*.py)plot_phy.py— path loss vs distance, BER vs Eb/N0plot_mac.py— Jain fairness index, HARQ rounds vs SNRplot_isac.py— DFRC Pareto frontier (CRB vs capacity, with Liu TSP 2018 overlay)plot_semantic.py— compression ratio vs task success, channel estimator NMSECI additions (
.github/workflows/ci.yml)matrix-features— builds and tests with"",baseline-comparison,plotting(three-way matrix)docker-build— buildsDockerfileon every PR; pushesghcr.io/j143/6g:latest+ SHA tag on merge tomain