Launch TeleMemetry Evaluation
Operational telemetry infrastructure for long-running AI systems
- One-click setup, cheap validation on an independent NVIDIA GPU
- You control turns, fields, and episodes
- Public harness boundary - inspect it before you share a claim
-OR-
We ran it so you don't have to
· download AI-ready analysis packages and verified logs
(.zip) or
(.tar.gz)
135,000
Boundary: this public reproduction harness validates the published scoring pipeline and artifact workflow. It does not expose or reproduce the private production engine or its proprietary implementation.
Video walkthrough
Video walkthrough
Walkthrough of the public GPU validation workflow: setup, first run, raw logs, and downloadable SHA256 analysis package.
What to expect
Visual walkthrough
Step-by-step screenshots
Hover any screenshot to zoom. Add your notes in the comment area below each image.
Before you launch
- Use Brev for quick validation on an independent NVIDIA GPU.
- You control the turn count, field count, and episode count before running.
- Delete any GPU environment you create when finished; provider billing can continue while instances run.
- Download raw logs and SHA256 analysis packages before deleting provider environments.
- The public reproduction harness is not the private production engine.
What this benchmark demonstrates
The public validation harness demonstrates the scoring pipeline, raw logs, and SHA256 artifact package for an independently run GPU validation pass.
It does not prove production TeleMemetry internals. It does not claim to replace all databases, RAG, compression, or pruning in all systems. The certified registry runs in the evidence library are produced by the private production harness and carry the full artifact chain required for engineering due diligence.
Run or inspect
Run the public harness
Open the Brev setup flow or inspect the reproduction repo first. The useful output is the package: raw input/output logs, scoring records, and SHA256 receipts you can check line by line.