Modelpublic

Yacinedh/translategemma-12b-khutbah-lora

sourceHugging Facegemmaupdated 3mo agoView on Hugging Face

0likes8downloads

Model Card

TranslateGemma-12B-Khutbah — LoRA adapter

The raw LoRA adapter (r=16, ~68M trainable parameters) from fine-tuning google/translategemma-12b-it on Islamic-sermon Arabic→German data.

Most users want the merged, ready-to-use model instead: [Yacinedh/translategemma-12b-khutbah](https://huggingface.co/Yacinedh/translategemma-12b-khutbah) (includes a Q4KM GGUF). This repo exists for reproducibility, further training, and adapter composition.

python

from peft import PeftModel
from transformers import AutoModelForCausalLM

base = AutoModelForCausalLM.from_pretrained("google/translategemma-12b-it", device_map="auto")
model = PeftModel.from_pretrained(base, "Yacinedh/translategemma-12b-khutbah-lora")

Training data, recipe, curves, and benchmarks: see the merged model card. Code: MinbarAI/training.