Update README.md
Browse files
README.md
CHANGED
|
@@ -2,12 +2,13 @@
|
|
| 2 |
license: apache-2.0
|
| 3 |
datasets:
|
| 4 |
- ramendik/kimify-ifeval-like
|
|
|
|
| 5 |
base_model:
|
| 6 |
- ibm-granite/granite-4.0-h-1b
|
| 7 |
---
|
| 8 |
|
| 9 |
Testing checkpoint of an attempt at distilling the style of Kimi K2 Instruct, inasmuch as possible in this size, by fine-tuning Granite 4.0 h 1b.
|
| 10 |
|
| 11 |
-
Long sequences made Granite prone to looping, so I made
|
| 12 |
|
| 13 |
This is clearly not the final checkpoint but I'd appreciate testing to see what needs to be improved.
|
|
|
|
| 2 |
license: apache-2.0
|
| 3 |
datasets:
|
| 4 |
- ramendik/kimify-ifeval-like
|
| 5 |
+
- ramendik/kimify-short-20260131
|
| 6 |
base_model:
|
| 7 |
- ibm-granite/granite-4.0-h-1b
|
| 8 |
---
|
| 9 |
|
| 10 |
Testing checkpoint of an attempt at distilling the style of Kimi K2 Instruct, inasmuch as possible in this size, by fine-tuning Granite 4.0 h 1b.
|
| 11 |
|
| 12 |
+
Long sequences made Granite prone to looping, so I made kimi-generated datasets system-prompted for brevity then filtered at max 1000t single asssistant response.
|
| 13 |
|
| 14 |
This is clearly not the final checkpoint but I'd appreciate testing to see what needs to be improved.
|