This is a LoRA adapter for TinyLlama (1.1B) fine-tuned using Unsloth optimizations (2x faster, 50% less memory). These are adapter weights only (~48 MB) that must be loaded on top of the base model.

🚀 Usage

python

# Install: pip install unsloth transformers peft accelerate

import torch
from unsloth import FastLanguageModel
from peft import PeftModel

# Load base model with Unsloth
BASE_MODEL = "TinyLlama/TinyLlama-1.1B-Chat-v1.0"
ADAPTER = "arif-butt/tinyllama-unsloth-lora-adapter"

base_model, tokenizer = FastLanguageModel.from_pretrained(
    model_name=BASE_MODEL,
    max_seq_length=512,
    dtype=torch.float16,
    device_map="auto",
)

# Attach adapter
model = PeftModel.from_pretrained(base_model, ADAPTER)
model.eval()

# Test prompt
prompt = "Q: Name all the courses Arif butt teach?\nA:"
inputs = tokenizer(prompt, return_tensors="pt").to(model.device)

with torch.no_grad():
    outputs = model.generate(
        **inputs,
        max_new_tokens=100,
        temperature=0.2,
        do_sample=True,
        pad_token_id=tokenizer.eos_token_id,
    )

response = tokenizer.decode(outputs[0], skip_special_tokens=True)
print(f"Response: {response}")