diff options
| author | skal <pascal.massimino@gmail.com> | 2026-02-13 12:49:05 +0100 |
|---|---|---|
| committer | skal <pascal.massimino@gmail.com> | 2026-02-13 12:49:05 +0100 |
| commit | aed21707f9ca43b70e7fbdae4144f9d64bd70d00 (patch) | |
| tree | e89836fc4e7aa5da84caa35c945d13ba8512b1a9 /doc/CNN_BIAS_FIX_2026-02.md | |
| parent | 5074e4caec017d6607de5806858d0271a554d77c (diff) | |
Doc: Clarify CNN v2 training uses RGBA targets
Updated CNN_V2.md to document that:
- Model outputs 4 channels (RGBA)
- Training targets preserve alpha from target images
- Loss function compares all 4 channels
Co-Authored-By: Claude Sonnet 4.5 <noreply@anthropic.com>
Diffstat (limited to 'doc/CNN_BIAS_FIX_2026-02.md')
0 files changed, 0 insertions, 0 deletions