Give each weight set its own version and commit
Browse files
README.md
CHANGED
|
@@ -24,7 +24,16 @@ Two weight sets are here. `stage2_best.pt` was trained on renders alone.
|
|
| 24 |
to see whether real data closed the gap. It did not, and both are published
|
| 25 |
rather than only the flattering one.
|
| 26 |
|
| 27 |
-
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 28 |
|
| 29 |
## The metric
|
| 30 |
|
|
|
|
| 24 |
to see whether real data closed the gap. It did not, and both are published
|
| 25 |
rather than only the flattering one.
|
| 26 |
|
| 27 |
+
| weights | version | commit |
|
| 28 |
+
|---|---|---|
|
| 29 |
+
| `stage2_best.pt` + `stage1_best.pt` | 263601 | `82d3030` |
|
| 30 |
+
| `v3mix_stage2.pt` | 263701 | `deed0ca` |
|
| 31 |
+
|
| 32 |
+
Both were briefly published under the same version number. The release counter
|
| 33 |
+
is derived from git tags, and with no tags it returns 01 forever, so two
|
| 34 |
+
different models were stamped 263701. Tags now exist for both, and the stamp
|
| 35 |
+
says UNTAGGED when none do rather than emitting a number that looks unique and
|
| 36 |
+
is not.
|
| 37 |
|
| 38 |
## The metric
|
| 39 |
|