29 lines
1.1 KiB
Markdown
29 lines
1.1 KiB
Markdown
|
# Issue #600 Verification: Visual Scenes Prompt Enhancement
|
||
|
|
|
||
|
|
**Status:** ✅ Complete — dataset present on main
|
||
|
|
**Issue:** [Timmy_Foundation/timmy-config#600](https://forge.alexanderwhitestone.com/Timmy_Foundation/timmy-config/issues/600)
|
||
|
|
**Dataset:** `training/data/prompt-enhancement/visual-scenes-500.jsonl`
|
||
|
|
**Records:** 500
|
||
|
|
**Domain:** `visual scenes` (all records)
|
||
|
|
|
||
|
|
## Validation
|
||
|
|
|
||
|
|
| Check | Result |
|
||
|
|
|-------|--------|
|
||
|
|
| File exists | ✅ |
|
||
|
|
| 500 JSONL records | ✅ |
|
||
|
|
| Valid JSON per line | ✅ |
|
||
|
|
| Required fields (terse, rich, domain) | ✅ |
|
||
|
|
| Domain equals "visual scenes" | ✅ |
|
||
|
|
| Non-empty terse and rich strings | ✅ |
|
||
|
|
| Full-record uniqueness | ✅ |
|
||
|
|
|
||
|
|
## Notes
|
||
|
|
|
||
|
|
- 65 terse prompts appear more than once (different rich expansions for same terse). The dataset contract specifies unique *pairs*, not unique terse prompts. Acceptable.
|
||
|
|
- File added via PR #731: `feat: 500 visual scene prompt enhancement pairs (#600)`. Merged to main.
|
||
|
|
|
||
|
|
## Files Added in This PR
|
||
|
|
|
||
|
|
- `tests/test_prompt_enhancement_visual_scenes.py` — regression test validating the dataset meets issue requirements.
|