Update README.md
Browse files
README.md
CHANGED
|
@@ -1,3 +1,25 @@
|
|
| 1 |
-
---
|
| 2 |
-
license: mit
|
| 3 |
-
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
---
|
| 2 |
+
license: mit
|
| 3 |
+
datasets:
|
| 4 |
+
- Vikhrmodels/GrandMaster-PRO-MAX
|
| 5 |
+
language:
|
| 6 |
+
- ru
|
| 7 |
+
- en
|
| 8 |
+
base_model:
|
| 9 |
+
- Qwen/Qwen2-VL-2B-Instruct
|
| 10 |
+
---
|
| 11 |
+
|
| 12 |
+
# tvl-mini
|
| 13 |
+
|
| 14 |
+
|
| 15 |
+
## Description
|
| 16 |
+
|
| 17 |
+
This is finetune of Qwen2-VL-2B on russian language.
|
| 18 |
+
|
| 19 |
+
tvl was trained in bf16
|
| 20 |
+
|
| 21 |
+
## Data
|
| 22 |
+
|
| 23 |
+
Train dataset contains:
|
| 24 |
+
- GrandMaster-PRO-MAX dataset (60k samples)
|
| 25 |
+
- Translated, humanized and merged by image subset of GQA (TODO)
|