Fix README with correct datasets/splits and curriculum notes
Browse files
README.md
CHANGED
|
@@ -15,17 +15,17 @@ datasets:
|
|
| 15 |
- murataksit34/veri-bilimci-diyalog-8k-tr
|
| 16 |
---
|
| 17 |
|
| 18 |
-
# Qwen2.5-Coder-3B-DataScience-TR
|
| 19 |
|
| 20 |
-
##
|
| 21 |
-
1.
|
| 22 |
-
2.
|
| 23 |
|
| 24 |
-
##
|
| 25 |
-
- Toplam
|
| 26 |
-
- Split:
|
| 27 |
-
- assistant_first_unique_ratio:
|
| 28 |
-
- assistant_final_unique_ratio:
|
| 29 |
|
| 30 |
## Copyright
|
| 31 |
Copyright (c) Zero9 Tech
|
|
|
|
| 15 |
- murataksit34/veri-bilimci-diyalog-8k-tr
|
| 16 |
---
|
| 17 |
|
| 18 |
+
# Qwen2.5-Coder-3B-DataScience-TR - Turkce Kodlu Analiz Asistani
|
| 19 |
|
| 20 |
+
## Egitim Kurgusu
|
| 21 |
+
1. Turkce dusunme adaptasyonu: wikimedia/wikipedia uzerinde yaklasik %10 on egitim/adaptasyon (48,148 kayit).
|
| 22 |
+
2. Uzmanlik SFT: murataksit34/veri-bilimci-diyalog-8k-tr.
|
| 23 |
|
| 24 |
+
## Kullanilan Dataset Ozeti
|
| 25 |
+
- Toplam kayit: 7,656
|
| 26 |
+
- Split: train 6,124 ; test 1,532
|
| 27 |
+
- assistant_first_unique_ratio: 0.7034
|
| 28 |
+
- assistant_final_unique_ratio: 0.8723
|
| 29 |
|
| 30 |
## Copyright
|
| 31 |
Copyright (c) Zero9 Tech
|