| # Submission Checklist |
|
|
| ## Required |
|
|
| - [x] Gradio Space created under `build-small-hackathon` |
| - [x] Space made public before June 15, 2026 |
| - [x] Demo video linked in Space README: https://youtu.be/oh0GRKYXvGM |
| - [x] Social post linked in Space README: https://www.reddit.com/r/huggingface/comments/1u48yt6/finetuned_a_1b_model_that_helps_families_check/ |
| - [x] Public GitHub repo linked in Space README |
| - [x] Repo includes Codex-attributed feature commits |
| - [x] README frontmatter includes track and target badge tags |
| - [x] MIT license file included |
| |
| ## Backyard AI Evidence |
| |
| - [x] Real user identified: friend's grandmother, private identity omitted |
| - [x] Real suspicious message pattern collected with private details removed |
| - [x] Privacy preserved: no public user quote is included; identity and private details are intentionally omitted |
| - [x] Demo script shows the real-user story |
| |
| ## Technical Evidence |
| |
| - [x] Model name and parameter count documented |
| - [x] Eval set included |
| - [x] Sanitized dataset page published on Hugging Face |
| - [x] Hugging Face collection links Space, model, dataset, and reports |
| - [x] Runtime decision documented |
| - [x] Local-first vs ZeroGPU demo boundary documented |
| - [x] Local inference path documented |
| - [x] Fine-tuned MiniCPM LoRA adapter published and evaluated |
| - [x] Modal training/eval evidence documented |
| - [x] No commercial cloud API path documented |
| |
| ## Bonus Badges |
| |
| - [x] Off-Brand |
| - [x] Well-Tuned |
| - [x] Sharing is Caring |
| - [x] Field Notes |
| - [x] Best Demo evidence finalized |
| |
| ## Codex Track Evidence |
| |
| - [x] Public GitHub repo exists |
| - [x] GitHub repo is linked from README |
| - [x] Codex build log exists |
| - [x] Agent trace exists |
| - [x] README includes file-level Codex contribution summary |
| - [x] Meaningful feature commits include Codex co-author trailer |
| - [x] UI warns users to redact sensitive private data before pasting |
| |
| ## Final Manual Tasks |
| |
| - [x] Set Space variables to MiniCPM5-1B + Jawbreaker LoRA v8. |
| - [x] Publish sanitized dataset/eval artifact page. |
| - [x] Create/update Hugging Face collection for the full submission package. |
| - [x] Choose one real or realistic sanitized scam message for the primary demo. |
| - [x] Publish Hugging Face article / community story: https://huggingface.co/blog/build-small-hackathon/jawbreaker-private-scam-defense |
| - [x] Record demo video. |
| - [x] Add demo video link to README. |
| - [x] Publish social post: https://www.reddit.com/r/huggingface/comments/1u48yt6/finetuned_a_1b_model_that_helps_families_check/ |
| - [x] Add social post link to README. |
| - [x] Make the Hugging Face Space public. |
| - [x] Final smoke test after making the Space public: `/health` returned `status=ok`, `backend=zerogpu`, and `model=MiniCPM5-1B + Jawbreaker LoRA v8` on June 12, 2026. |
|
|
