Publish s6-micro-v1
Browse files
README.md
CHANGED
|
@@ -144,8 +144,8 @@ Face, classic datasets converted into typed questions, game states labelled by e
|
|
| 144 |
numeric curriculum, the same as s6-nano. Pretraining text: SimpleStories, TinyStories, FineWeb and
|
| 145 |
the non-web parts of Dolma 3 and TxT360. Soft labels on 747,077 training questions come from
|
| 146 |
GLM-5.3-flash (open weights, MIT, called through the z.ai API), kept only where they agree with
|
| 147 |
-
the gold answer; no test question was sent. The
|
| 148 |
-
|
| 149 |
[DATA_LICENSES.md](https://github.com/maxgfr/skycmd/blob/main/DATA_LICENSES.md), with every
|
| 150 |
other license.
|
| 151 |
|
|
|
|
| 144 |
numeric curriculum, the same as s6-nano. Pretraining text: SimpleStories, TinyStories, FineWeb and
|
| 145 |
the non-web parts of Dolma 3 and TxT360. Soft labels on 747,077 training questions come from
|
| 146 |
GLM-5.3-flash (open weights, MIT, called through the z.ai API), kept only where they agree with
|
| 147 |
+
the gold answer; no test question was sent. The teacher's weights are open (MIT); the labels and
|
| 148 |
+
their publication are described in
|
| 149 |
[DATA_LICENSES.md](https://github.com/maxgfr/skycmd/blob/main/DATA_LICENSES.md), with every
|
| 150 |
other license.
|
| 151 |
|