Create README.md
Browse files
README.md
ADDED
|
@@ -0,0 +1,24 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
---
|
| 2 |
+
base_model:
|
| 3 |
+
- google/gemma-3-12b-it
|
| 4 |
+
---
|
| 5 |
+
|
| 6 |
+
About 13.5M tokens total of mixed instruct and RP data.
|
| 7 |
+
|
| 8 |
+
Both RP datasets and the inkstruct include system prompts to help g3 understand the system role (via `<start_of_turn>system`).
|
| 9 |
+
|
| 10 |
+
```yaml
|
| 11 |
+
datasets:
|
| 12 |
+
- path: ToastyPigeon/some-rp-extended
|
| 13 |
+
type: customgemma-regex
|
| 14 |
+
- path: allura-org/inkstructmix-v0.2.1a-system-reasoning-separated
|
| 15 |
+
type: customgemma-regex
|
| 16 |
+
data_files: inkstruct-system.json
|
| 17 |
+
split: train[:750]
|
| 18 |
+
- path: ToastyPigeon/unalign-v2
|
| 19 |
+
type: customgemma-regex
|
| 20 |
+
split: train[:50%]
|
| 21 |
+
- path: ToastyPigeon/synth-rp
|
| 22 |
+
split: train[:20%]
|
| 23 |
+
type: customgemma-regex
|
| 24 |
+
```
|