Evaluating Anthropic models

Config

Ensure you have the following config in your config.json file:

{
  "models": [
    {
      "provider": "Anthropic",
      "model": "claude-3-5-sonnet-20240620",
      "model_name": "Claude 3.5 Sonnet",
      "endpoint": "https://api.anthropic.com/v1/messages",
      "api_key_env_var": "CLAUDE_API_KEY"
    },
    {
      "provider": "Anthropic",
      "model": "claude-3-opus-20240229",
      "model_name": "Claude 3 Opus",
      "endpoint": "https://api.anthropic.com/v1/messages",
      "api_key_env_var": "CLAUDE_API_KEY"
    },
    {
      "provider": "Anthropic",
      "model": "claude-3-haiku-20240307",
      "model_name": "Claude 3 Haiku",
      "endpoint": "https://api.anthropic.com/v1/messages",
      "api_key_env_var": "CLAUDE_API_KEY"
    }
  ],
  "global_parameters": {
    "temperature": 1,
    "max_tokens": 200,
    "top_p": 1,
    "frequency_penalty": 0,
    "presence_penalty": 0
  }
}

Ensure you have the following config in your .env file:

CLAUDE_API_KEY=your_api_key

Run evaluate

thabit eval --dataset-name=unicode-text --models=claude-3-5-sonnet-20240620,claude-3-opus-20240229,claude-3-haiku-20240307