Quantization made by Richard Erkhov.

Fine-Tuning-Gemma-2b-it-for-Arabic - GGUF

Model creator: https://huggingface.co./Ruqiya/
Original model: https://huggingface.co./Ruqiya/Fine-Tuning-Gemma-2b-it-for-Arabic/

Name	Quant method	Size
Fine-Tuning-Gemma-2b-it-for-Arabic.Q2_K.gguf	Q2_K	1.08GB
Fine-Tuning-Gemma-2b-it-for-Arabic.IQ3_XS.gguf	IQ3_XS	1.16GB
Fine-Tuning-Gemma-2b-it-for-Arabic.IQ3_S.gguf	IQ3_S	1.2GB
Fine-Tuning-Gemma-2b-it-for-Arabic.Q3_K_S.gguf	Q3_K_S	1.2GB
Fine-Tuning-Gemma-2b-it-for-Arabic.IQ3_M.gguf	IQ3_M	1.22GB
Fine-Tuning-Gemma-2b-it-for-Arabic.Q3_K.gguf	Q3_K	1.29GB
Fine-Tuning-Gemma-2b-it-for-Arabic.Q3_K_M.gguf	Q3_K_M	1.29GB
Fine-Tuning-Gemma-2b-it-for-Arabic.Q3_K_L.gguf	Q3_K_L	1.36GB
Fine-Tuning-Gemma-2b-it-for-Arabic.IQ4_XS.gguf	IQ4_XS	1.4GB
Fine-Tuning-Gemma-2b-it-for-Arabic.Q4_0.gguf	Q4_0	1.44GB
Fine-Tuning-Gemma-2b-it-for-Arabic.IQ4_NL.gguf	IQ4_NL	1.45GB
Fine-Tuning-Gemma-2b-it-for-Arabic.Q4_K_S.gguf	Q4_K_S	1.45GB
Fine-Tuning-Gemma-2b-it-for-Arabic.Q4_K.gguf	Q4_K	1.52GB
Fine-Tuning-Gemma-2b-it-for-Arabic.Q4_K_M.gguf	Q4_K_M	1.52GB
Fine-Tuning-Gemma-2b-it-for-Arabic.Q4_1.gguf	Q4_1	1.56GB
Fine-Tuning-Gemma-2b-it-for-Arabic.Q5_0.gguf	Q5_0	1.68GB
Fine-Tuning-Gemma-2b-it-for-Arabic.Q5_K_S.gguf	Q5_K_S	1.68GB
Fine-Tuning-Gemma-2b-it-for-Arabic.Q5_K.gguf	Q5_K	1.71GB
Fine-Tuning-Gemma-2b-it-for-Arabic.Q5_K_M.gguf	Q5_K_M	1.71GB
Fine-Tuning-Gemma-2b-it-for-Arabic.Q5_1.gguf	Q5_1	1.79GB
Fine-Tuning-Gemma-2b-it-for-Arabic.Q6_K.gguf	Q6_K	1.92GB
Fine-Tuning-Gemma-2b-it-for-Arabic.Q8_0.gguf	Q8_0	2.49GB

Original model description:

datasets: - arbml/CIDAR base_model: google/gemma-2b-it pipeline_tag: text-generation language: - ar - en

Fine-Tuning-Gemma-2b-it-for-Arabic

This model is a fine-tuned version of google/gemma-2b-it on arbml/CIDAR Arabic dataset.

It achieves the following results on the evaluation set:

training_loss=2.281057505607605

Direct Use

!pip install -qU transformers accelerate

from transformers import AutoTokenizer
import transformers
import torch

model = "Ruqiya/Fine-Tuning-Gemma-2b-it-for-Arabic"
messages = [{"role": "user", "content": "ما هو الذكاء الاصطناعي؟"}]

tokenizer = AutoTokenizer.from_pretrained(model)
prompt = tokenizer.apply_chat_template(messages, tokenize=False, add_generation_prompt=True)
pipeline = transformers.pipeline(
    "text-generation",
    model=model,
    torch_dtype=torch.float16,
    device_map="auto",
)

outputs = pipeline(prompt, max_new_tokens=256, do_sample=True, temperature=0.7, top_k=50, top_p=0.95)
print(outputs[0]["generated_text"])