promptfoo/examples/deepseek-r1-vs-openai-o1

deepseek-r1-vs-openai-o1 (DeepSeek-R1 vs OpenAI o1 Comparison)

You can run this example with:

npx promptfoo@latest init --example deepseek-r1-vs-openai-o1

This example demonstrates how to benchmark DeepSeek's R1 model against OpenAI's o1 model using the Massive Multitask Language Understanding (MMLU) benchmark, focusing on reasoning-heavy subjects.

Prerequisites

• promptfoo CLI installed (npm install -g promptfoo or brew install promptfoo)
• OpenAI API key set as OPENAI_API_KEY
• DeepSeek API key set as DEEPSEEK_API_KEY
• Hugging Face account and access token (for MMLU dataset)

Hugging Face Authentication

To access the MMLU dataset, you'll need to authenticate with Hugging Face:

1. Create a Hugging Face account at huggingface.co if you don't have one

2. Generate an access token at huggingface.co/settings/tokens

3. Set your token as an environment variable:

   export HF_TOKEN=your_token_here
   

   Or add it to your .env file:

   HF_TOKEN=your_token_here
   

Running the Eval

1. Get a local copy of the promptfooconfig.

   You can clone this repository and from the root directory run:

   cd examples/deepseek-r1-vs-openai-o1
   

   or you can get the example with:

   promptfoo init --example deepseek-r1-vs-openai-o1
   

2. Run the evaluation:

   promptfoo eval
   

3. View the results in a web interface:

   promptfoo view
   

What's Being Tested

This comparison evaluates both models on reasoning tasks from the MMLU benchmark, specifically:

1. Abstract Algebra: Advanced mathematical reasoning
2. Formal Logic: Logical statement analysis
3. High School Mathematics: Core problem-solving
4. College Mathematics: Advanced mathematical concepts
5. Logical Fallacies: Flaw identification in reasoning

Each subject uses 10 questions to keep the test manageable. You can edit this in promptfooconfig.yaml.

Test Structure

The configuration in promptfooconfig.yaml:

1. Prompt Template: Encourages step-by-step reasoning for multiple choice questions
2. Quality Checks:
   • 60-second timeout per question
   • Required step-by-step reasoning
   • Clear final answer format
3. Evaluation Metrics:
   • Accuracy
   • Reasoning quality
   • Response time
   • Format adherence

Customizing

You can modify the test by editing promptfooconfig.yaml:

1. Add more MMLU subjects:

   tests:
     - huggingface://datasets/cais/mmlu?split=test&subset=physics
   

2. Try different prompting strategies:

   prompts:
     # Zero-shot with step-by-step reasoning (default)
     - |
       You are an expert test taker. Please solve the following multiple choice question step by step.
   
       Question: {{question}}
   
       Options:
       A) {{choices[0]}}
       B) {{choices[1]}}
       C) {{choices[2]}}
       D) {{choices[3]}}
   
       Think through this step by step, then provide your final answer in the format "Therefore, the answer is A/B/C/D."
   
     # Zero-shot with direct answer
     - |
       Question: {{question}}
   
       A) {{choices[0]}}
       B) {{choices[1]}}
       C) {{choices[2]}}
       D) {{choices[3]}}
   
       Answer with just the letter (A/B/C/D) of the correct option.
   

3. Change the number of questions:

   tests:
     - huggingface://datasets/cais/mmlu?split=test&subset=physics&limit=20 # Test 20 questions per subject
   

4. Adjust quality requirements:

   defaultTest:
     assert:
       - type: latency
         threshold: 30000 # Stricter 30-second timeout
   

Additional Resources

• DeepSeek provider documentation
• MMLU benchmark