Submission/dinuda

[Dinuda]

MIB Doc Challenge Submission

Thanks for submitting! Two required steps:

1. Fill out the submission form: https://docs.google.com/forms/d/1ZLkHmTsYd9I87JL1sUyps2rPTe6ohEI_lTZ8Jjts6bw/viewform
   Your entry does not count until the form is submitted.
2. Complete this pull request. It should only add files under submissions/<your-github-username>/.

Links

• Solution repository (public, contains a Dockerfile):
• Submission form: I filled it out on (date):

Checklist

• [√ ] I filled out the submission form linked above
• [ √] This PR only adds submissions/<my-github-username>/predictions.jsonl, MEMO.md, and SUBMISSION.md
• [ √] predictions.jsonl passes scripts/validate_submission.py against data/validation_manifest.csv
• [ √] My solution repository is public and includes a Dockerfile
• [ √] My Docker image runs offline (--network none) and accepts <input_pdf_dir> <output_predictions_path>
• [ √] My submitted runtime uses no LLMs, VLMs, cloud OCR, or network services
• [ √] Model artifacts fit the size limits in DOCKER_SUBMISSION.md
• [√ ] No hardcoded validation answers and no manual per-case edits
• [√ ] My memo describes my approach, failure modes, and what I would improve with another week

[Copilot]

Pull request overview

Adds Dinuda’s MIB Doc Challenge submission under submissions/Dinuda/, providing the required validation predictions plus accompanying write-ups describing the approach and how to run the offline Docker solution.

Changes:

• Added predictions.jsonl containing 5,000 validation-set predictions.
• Added MEMO.md technical memo describing the offline pipeline, training-label usage constraints, and failure modes.
• Added SUBMISSION.md with the public solution repository link and Docker run instructions.

Reviewed changes

Copilot reviewed 2 out of 3 changed files in this pull request and generated 2 comments.

File	Description
submissions/Dinuda/predictions.jsonl	5,000-line JSONL of validation predictions for scoring.
submissions/Dinuda/MEMO.md	Technical memo documenting approach, constraints, and analysis.
submissions/Dinuda/SUBMISSION.md	Submission summary, repo link, and runtime contract/run instructions.

Suppressed comments (2)

submissions/Dinuda/MEMO.md:123

• "two small logistics" reads like a typo; this section is describing two small logistic regression models. Consider using "logistic regressions" (or "logistic models") to avoid confusion with general operational logistics.

   unreadable). Structural packet features only — page roles, damage, coverage,
   that kind of thing — go into two small logistics. Both have to fire before
   the row may flip to `DENIED`. It cannot approve, and it cannot see identity

submissions/Dinuda/MEMO.md:126

• This appears to refer to a logistic regression model; "another small logistic" is incomplete/unclear terminology.

2. **Path calibrator.** After the decision is frozen, another small logistic
   maps eight structural features to confidence. It does not change the action.

---

💡 Add Copilot custom instructions for smarter, more guided reviews. Learn how to get started.