gunbaz/twin-llama-3.1-8b

Yaşar Kemal Digital Twin - Llama 3.1 8B LoRA

This model is a fine-tuned version of meta-llama/Meta-Llama-3.1-8B-Instruct using LoRA (Low-Rank Adaptation).

Model Description

• Base Model: Llama 3.1 8B Instruct
• Fine-tuning Method: LoRA (r=16, alpha=16)
• Quantization: 4-bit (QLoRA)
• Training Data: Yaşar Kemal literary works
• Training Library: Unsloth + PEFT

Intended Use

This model is designed to generate text in the style of Yaşar Kemal, the renowned Turkish novelist.

Usage

from peft import AutoPeftModelForCausalLM
from transformers import AutoTokenizer

model = AutoPeftModelForCausalLM.from_pretrained(
    "gunbaz/twin-llama-3.1-8b",
    load_in_4bit=True,
)
tokenizer = AutoTokenizer.from_pretrained("gunbaz/twin-llama-3.1-8b")

# Generate
prompt = "Yaşar Kemal'in üslubuyla doğayı anlat."
inputs = tokenizer(prompt, return_tensors="pt").to("cuda")
outputs = model.generate(**inputs, max_new_tokens=128)
print(tokenizer.decode(outputs[0], skip_special_tokens=True))

Training Details

• Dataset Size: 4,862 instruction-response pairs
• LoRA Rank: 16
• Training Framework: Unsloth
• Optimizer: AdamW 8-bit
• Learning Rate: 2e-4

License

Apache 2.0 (following base model license)