from transformers import AutoTokenizer, AutoModelForCausalLM
from peft import PeftModel, PeftConfig
import torch

base_model_name = "Haleshot/Mathmate-7B-DELLA"
adapter_name = "Haleshot/Mathmate-7B-DELLA-ORPO-C"

base_model = AutoModelForCausalLM.from_pretrained(base_model_name, torch_dtype=torch.float16, device_map="auto")
tokenizer = AutoTokenizer.from_pretrained(base_model_name)
model = PeftModel.from_pretrained(base_model, adapter_name)

def generate_response(prompt, max_length=512):
    inputs = tokenizer(prompt, return_tensors="pt").to(model.device)
    outputs = model.generate(**inputs, max_length=max_length, num_return_sequences=1, do_sample=True, temperature=0.7)
    return tokenizer.decode(outputs[0], skip_special_tokens=True)

prompt = "Let's have a casual conversation about the weather today."
response = generate_response(prompt)
print(response)

Acknowledgements

Base Model: <a href="https://huggingface.co/Haleshot/Mathmate-7B-DELLA-ORPO" target="_blank" rel="noopener noreferrer" class="text-blue-400 hover:text-blue-300 underline underline-offset-2 transition-colors">Haleshot/Mathmate-7B-DELLA
Training Dataset: <a href="https://huggingface.co/datasets/HuggingFaceTB/everyday-conversations-llama3.1-2k" target="_blank" rel="noopener noreferrer" class="text-blue-400 hover:text-blue-300 underline underline-offset-2 transition-colors">HuggingFaceTB/everyday-conversations-llama3.1-2k

Thanks to the HuggingFaceTB team for providing the everyday conversations dataset used in this finetuning process.