Downloads · 30 days
21
11% of all-time downloads
Yvthyvq/Liujgoj-Cantonese-Gemma4-12b-Base
Liujgoj-Cantonese-Gemma4-12b-Base is a text generation model from Yvthyvq. Use it when you need the model to write or continue text. The card lists the license as apache-2.0.
Liujgoj-Cantonese-Gemma4-12b-Base 是一個基於 Gemma 4 12B 進行擴充詞表與增量預訓練(Continued Pre-training, CPT)的粵語大語言模型基座。
Downloads · 30 days
21
11% of all-time downloads
All-time downloads
194
Public
Parameters
12B
24 GB on disk
Likes
1
Public
Click a slice to open those files.
.safetensors24 GB · 100%
From the Hugging Face model README
Liujgoj-Cantonese-Gemma4-12b-Base 是一個基於 Gemma 4 12B 進行擴充詞表與增量預訓練(Continued Pre-training, CPT)的粵語大語言模型基座。
本模型的核心特色是深度集成了 Liujgoj 粵語(Liujgoj Cantonese) —— 一套專為粵語設計的實驗性、基於拉丁字母的專屬正詞法/拼寫系統(Latin-based orthography)。我們在基座模型中預先擴充並初始化了 6,000 個 Liujgoj 專屬音素與詞條,使模型能夠原生、無字符障礙地進行高效率的廣東話語意建模與文本生成。
GGUF Version available here: https://huggingface.co/Yvthyvq/Liujgoj-Cantonese-Gemma4-12b-Base-Q4_K_M-GGUF
本模型在增量預訓練(CPT)部分使用了 OpenCSG/Fineweb-Edu-Chinese 數據集。在此特別感謝 OpenCSG 開源社區提供的優質教育語料。本模型的衍生、分發與使用將嚴格遵循 OpenCSG 模型社區許可協議。
Part of the continued pre-training (CPT) data used in this model is derived from the HuggingFaceFW/fineweb-edu dataset. We are grateful to the Hugging Face vLLM/FW team for providing this high-quality dataset under the MIT license.
你可以使用 transformers 庫直接載入並調用本模型:
import torch
from transformers import AutoModelForCausalLM, AutoTokenizer
model_id = "Yvthyvq/Liujgoj-Cantonese-Gemma4-12b-Base"
tokenizer = AutoTokenizer.from_pretrained(model_id, trust_remote_code=True)
model = AutoModelForCausalLM.from_pretrained(
model_id,
torch_dtype=torch.bfloat16,
device_map="auto",
trust_remote_code=True
)
# 測試你的 Liujgoj 粵語輸入
text = "請用 Liujgoj 粵語輸入你的 Prompt..."
inputs = tokenizer(text, return_tensors="pt").to(model.device)
with torch.no_grad():
outputs = model.generate(**inputs, max_new_tokens=512)
print(tokenizer.decode(outputs[0], skip_special_tokens=True))