Update README.md
Browse files
README.md
CHANGED
|
@@ -9,6 +9,8 @@ tags:
|
|
| 9 |
- qwen2
|
| 10 |
- trl
|
| 11 |
- sft
|
|
|
|
|
|
|
| 12 |
---
|
| 13 |
<!DOCTYPE html>
|
| 14 |
<html lang="en">
|
|
@@ -39,11 +41,13 @@ Bakpia V1 is a family of Javanese language models. It is fine-tuned from availab
|
|
| 39 |
|
| 40 |
This repository contains the fp16 version of Bakpia V1 1.5B.
|
| 41 |
|
| 42 |
-
| Version | Base Model | URL |
|
| 43 |
-
|
| 44 |
-
| V1 0.5B | Qwen 2 0.5B Instruct | [fp16](huggingface.co/afrizalha/Bakpia-V1-0.5B-Javanese/) |
|
| 45 |
-
| V1 1.5B | Qwen 2 1.5B Instruct | [fp16](huggingface.co/afrizalha/Bakpia-V1-1.5B-Javanese/) |
|
| 46 |
-
| V1 9B | Gemma 2 9B Instruct | [fp16](huggingface.co/afrizalha/Bakpia-V1-9B-Javanese-fp16)/[4bit](huggingface.co/afrizalha/Bakpia-V1-9B-Javanese-4bit/) |
|
|
|
|
|
|
|
| 47 |
|
| 48 |
## Version 1.0
|
| 49 |
|
|
|
|
| 9 |
- qwen2
|
| 10 |
- trl
|
| 11 |
- sft
|
| 12 |
+
datasets:
|
| 13 |
+
- afrizalha/Gatra-2-Javanese
|
| 14 |
---
|
| 15 |
<!DOCTYPE html>
|
| 16 |
<html lang="en">
|
|
|
|
| 41 |
|
| 42 |
This repository contains the fp16 version of Bakpia V1 1.5B.
|
| 43 |
|
| 44 |
+
| Version | Base Model | URL | Training |
|
| 45 |
+
|---------|------------|-----|----------|
|
| 46 |
+
| V1 0.5B | Qwen 2 0.5B Instruct | [fp16](huggingface.co/afrizalha/Bakpia-V1-0.5B-Javanese/) | Epoch = 1, Batch = 16\*8, lr = 5e-5, linear schedule|
|
| 47 |
+
| V1 1.5B | Qwen 2 1.5B Instruct | [fp16](huggingface.co/afrizalha/Bakpia-V1-1.5B-Javanese/) | Epoch = 1, Batch = 16\*8, lr = 5e-5, linear schedule|
|
| 48 |
+
| V1 9B | Gemma 2 9B Instruct | [fp16](huggingface.co/afrizalha/Bakpia-V1-9B-Javanese-fp16)/[4bit](huggingface.co/afrizalha/Bakpia-V1-9B-Javanese-4bit/) |Batch size = 16\*8, lr = 4e-5, linear schedule|
|
| 49 |
+
|
| 50 |
+
Training data is accessible here: [URL](https://huggingface.co/datasets/afrizalha/Gatra-2-Javanese)
|
| 51 |
|
| 52 |
## Version 1.0
|
| 53 |
|