ParlzStar X
ParlzStar X ๆฏไธไธชๅบไบ Qwen3-4B ๅพฎ่ฐ็ไปฃ็ ๅฉๆ๏ผ็ฑ JGZ_YES ่ฎญ็ปใ
ๆจกๅไฟกๆฏ
| ้กน็ฎ | ๅ ๅฎน |
|---|---|
| ๅบๅบงๆจกๅ | Qwen/Qwen3-4B |
| ่ฎญ็ปๆนๆณ | QLoRA (r=16, alpha=16) |
| ่ฎญ็ปๆฐๆฎ | Chinese-Instruct-Lite code ๅญ้ (3000ๆก) + ๅๆ่บซไปฝๆฐๆฎ (1000ๆก) |
| ่ฎญ็ปๆญฅๆฐ | 800 ๆญฅ |
| ๆ็ป loss | 0.4154 |
| ่ฎญ็ป็กฌไปถ | RTX 5060 8GB |
| ่ฎญ็ป่ | JGZ_YES |
่ฝๅ
- ไปฃ็ ็ๆ๏ผPythonใC++ใJavaใRustใJavaScript ็ญ
- ็ฎๆณๅฎ็ฐ๏ผๆๅบใๆฅๆพใๅจๆ่งๅใๆฐๆฎ็ปๆ
- ไปฃ็ ่งฃ้ไธ่ฐ่ฏ
- ่บซไปฝ่ฎค็ฅ๏ผ็ฅ้่ชๅทฑๆฏ ParlzStar X๏ผ็ฑ JGZ_YES ่ฎญ็ป
ไฝฟ็จๆนๅผ
ๅ ่ฝฝ LoRA ้้ ๅจ๏ผ
from unsloth import FastLanguageModel
model, tokenizer = FastLanguageModel.from_pretrained(
model_name="JGZYES/ParlzStar-X",
max_seq_length=2048,
load_in_4bit=True,
)
FastLanguageModel.for_inference(model)
ไฝฟ็จ GGUF๏ผ
llama-cli -m parlzstar-x-q4km.gguf -c 4096 -ngl 99 --repeat-penalty 1.3
ๆไปถ่ฏดๆ
| ๆไปถ | ่ฏดๆ |
|---|---|
| adapter_model.safetensors | LoRA ๆ้ |
| adapter_config.json | LoRA ้ ็ฝฎ |
| parlzstar-x-q4km.gguf | Q4_K_M ้ๅ๏ผ็บฆ 2.4GB |
| parlzstar-x-q6k.gguf | Q6_K ้ๅ๏ผ็บฆ 3.1GB |
| parlzstar-x-q8.gguf | Q8_0 ้ๅ๏ผ็บฆ 4.0GB |
ๅฑ้ๆง
4B ๅๆฐ่งๆจก๏ผๅคๆ็ฎๆณๆจ็่ฝๅๆ้
็ๆ็ไปฃ็ ่ฏทๅกๅฟ ้ช่ฏๅๅไฝฟ็จ
ๅถๅฐไผ่พๅบ็นๆฎๆ ็ญพ๏ผๆจ็ๆถๅฏ็จ stop_strings ่ฟๆปค
่บซไปฝ่ฎค็ฅๅถๅฐๅญๅจๆณขๅจ
่ฎธๅฏ่ฏ
ๆฌๆจกๅๅบไบ Qwen/Qwen3-4B ่ฎญ็ปใ่ฎญ็ปๆฐๆฎ Chinese-Instruct-Lite ไฝฟ็จ CC BY-NC-SA 4.0 ่ฎธๅฏ่ฏ๏ผๅ ๆญคๆฌๆจกๅๅๆ ท้็จ CC BY-NC-SA 4.0๏ผ็ฆๆญขๅไธ็จ้ใ
ๅผ็จ
@misc{parlzstar-x,
title={ParlzStar X},
author={JGZ_YES},
year={2025},
url={https://huggingface.co/JGZYES/ParlzStar-X}
}
- Downloads last month
- 435
Hardware compatibility
Log In to add your hardware
We're not able to determine the quantization variants.
Inference Providers NEW
This model isn't deployed by any Inference Provider. ๐ Ask for provider support