Work with different LLMs and parameters

A Python implementation task to work with different LLMs (Large Language Models) and request parameters via DIAL API

🎓 Learning Goals

By completing these tasks, you will learn:

• How to work with different LLMs through DIAL API
• How can we configure LLM output via request parameters (temperature, n, seed, etc...)

📋 Requirements

• Python 3.11+
• pip
• API key for DIAL service
• Basic understanding of HTTP requests and async/await

🔧 Setup

1. Install dependencies:

   pip install -r requirements.txt

2. Set your API key:

   • Ensure that you connected to the EPAM VPN
   • Get the DIAL API key here: https://support.epam.com/ess?id=sc_cat_item&table=sc_cat_item&sys_id=910603f1c3789e907509583bb001310c
   • Get available models from: https://ai-proxy.lab.epam.com/openai/models

3. Project structure:

   task/
   ├── models/
   │   ├── conversation.py          ✅ Complete
   │   ├── message.py               ✅ Complete  
   │   └── role.py                  ✅ Complete   
   ├── app/
   │   ├── main.py                  ✅ Complete
   │   └── client.py                ✅ Complete
   ├── 1-task-models.py             🚧 TODO
   ├── 2-task-n.py                  🚧 TODO
   ├── 3-task-temperature.py        🚧 TODO
   ├── 4-task-seed.py               🚧 TODO
   ├── 5-task-max_tokens.py         🚧 TODO
   ├── 6-task-frequency_penalty.py  🚧 TODO
   ├── 7-task-presence_penalty.py   🚧 TODO
   └── 8-task-stop.py               🚧 TODO
   

Available models:

• gpt-4o
• gpt-4.1-nano-2025-04-14
• gpt-4.1-mini-2025-04-14
• gemini-2.0-flash-lite
• gemini-2.0-flash
• gemini-2.5-pro
• gemini-2.5-flash
• claude-3-5-haiku@20241022
• claude-3-7-sonnet@20250219
• claude-sonnet-4@20250514

📝 Your Tasks

Implement all tasks from these files:

• 1-task-models.py
• 2-task-n.py
• 3-task-temperature.py
• 4-task-seed.py
• 5-task-max_tokens.py
• 6-task-frequency_penalty.py
• 7-task-presence_penalty.py
• 8-task-stop.py

💪 Additional task:

Practice with other parameters from OpenAI and Anthropic. For instance OpenAI have reasoning_effort and Anthropic thinking, and there are many others like citations, etc...

Pay attention that we are using DIAL Unified protol and all parameters that are not present in here https://dialx.ai/dial_api#operation/sendChatCompletionRequest must be provided as {"custom_fields": {"configuration": {CUSTOM_PARAMETERS} }}. More about custom_fields read here https://dialx.ai/dial_api#operation/sendChatCompletionRequest it is the last parameter described in documentation!

---