README.md

metadata

license: apache-2.0

🧩 Simple Use Case

This section demonstrates a simple use case of how to interact with our model to solve problems in a step-by-step, friendly manner.

Define the Function

We define a function get_completion which takes user input, combines it with a predefined system prompt, and then sends this combined prompt to our model. The model's response is then printed out.

Here's how the function is implemented:

import torch                
from transformers import pipeline   
import os

# Load model
test_pipeline = pipeline(model="zaursamedov1/FIxtral",
                            torch_dtype=torch.bfloat16,
                            trust_remote_code=True,
                            device_map="auto")

### Define the function
def get_completion(input):
    system = "Think step by step and solve the problem in a friendly way."
    prompt = f"#### System: {system}\\n#### User: \\n{input}\\n\\n#### Response from FIxtral model:"
    print(prompt)
    fixtral_prompt = test_pipeline(prompt, max_new_tokens=500)
    return fixtral_prompt[0]["generated_text"]

# Let's prompt
prompt = "problem"
print(get_completion(prompt))