chatgpt-cli/docs/chatgpt_api.md

ChatGPT API

OpenAI

Documentation

https://platform.openai.com/docs/api-reference/chat/create

CURLS

Request:

curl --location --insecure --request POST 'https://api.openai.com/v1/chat/completions' \
  --header "Authorization: Bearer ${OPENAI_API_KEY}" \
  --header 'Content-Type: application/json' \
  --data-raw '{
     "model": "gpt-3.5-turbo",
     "messages": [{"role": "user", "content": "What is the OpenAI mission?"}],
     "stream": false
  }' | jq .

Response:

{
  "id": "chatcmpl-7BQnIwmXhD6ohmwS6BjRHJrw9rJ7K",
  "object": "chat.completion",
  "created": 1682958960,
  "model": "gpt-3.5-turbo-0301",
  "usage": {
    "prompt_tokens": 15,
    "completion_tokens": 96,
    "total_tokens": 111
  },
  "choices": [
    {
      "message": {
        "role": "assistant",
        "content": "The OpenAI mission is to create and develop advanced Artificial Intelligence in a way that is safe and beneficial to humanity. Their goal is to build systems capable of performing tasks that would normally require human intelligence, such as natural language processing, understanding, and decision-making. The organization aims to foster research and development that is accessible and open to the public while maintaining ethical considerations and prioritizing safety. The ultimate objective is to use AI to enhance human life while minimizing the potential for negative consequences."
      },
      "finish_reason": "stop",
      "index": 0
    }
  ]
}

Or flip stream to true (this results in retrieving a ton of jsonl).

curl --location --insecure --request POST 'https://api.openai.com/v1/chat/completions' \
  --header "Authorization: Bearer ${OPENAI_API_KEY}" \
  --header 'Content-Type: application/json' \
  --data-raw '{
     "model": "gpt-3.5-turbo",
     "messages": [{"role": "user", "content": "What is the OpenAI mission?"}],
     "stream": true
  }'

... top omitted ... 

data: {"id":"chatcmpl-8B1ELWT5QKYmUbH0Az9anpvoOVdGZ","object":"chat.completion.chunk","created":1697637029,"model":"gpt-3.5-turbo-0613","choices":[{"index":0,"delta":{"content":" power"},"finish_reason":null}]}

data: {"id":"chatcmpl-8B1ELWT5QKYmUbH0Az9anpvoOVdGZ","object":"chat.completion.chunk","created":1697637029,"model":"gpt-3.5-turbo-0613","choices":[{"index":0,"delta":{"content":"."},"finish_reason":null}]}

data: {"id":"chatcmpl-8B1ELWT5QKYmUbH0Az9anpvoOVdGZ","object":"chat.completion.chunk","created":1697637029,"model":"gpt-3.5-turbo-0613","choices":[{"index":0,"delta":{},"finish_reason":"stop"}]}

data: [DONE]

Providing custom context

You can provide your own context in the messages array in your callout. You can split this data over multiple lines. For example

[
  {
    "role": "system",
    "content": "You are a helpful assistant."
  },
  {
    "role": "user",
    "content": "My dog's name is Piet Kernbom"
  },
  {
    "role": "user",
    "content": "and he's a 3-year-old Golden Retriever. He loves playing fetch and going for long walks."
  }
]

Using the chatgpt-cli with this context:

./bin/chatgpt What kind of toys would be best for Piet Kernbom?
Golden Retrievers like Piet Kernbom typically enjoy playing with toys that can be used for interactive play and also provide mental stimulation. Some good options for him may include:

1. Tennis Balls: Since Piet loves playing fetch, tennis balls can be great for him as they can be thrown over long distances.

2. Rope Toys: Rope toys are great for interactive play and also help with dental health. They come in a variety of shapes and sizes that Piet can choose from.

3. Puzzle Toys: Puzzle toys like treat balls are great for providing mental stimulation and keeping Piet entertained while you're away.

4. Soft Toys: Soft toys make great comfort toys for Golden Retrievers and can provide them with a sense of security.

Overall, it's good to keep a variety of toys on hand and observe your dog's preferences to find out what he likes best.

List Models

curl https://api.openai.com/v1/models \
  -H "Authorization: Bearer ${OPENAI_API_KEY}"

curl DALL-E

curl https://api.openai.com/v1/images/generations \
  -H "Content-Type: application/json" \
  -H "Authorization: Bearer ${OPENAI_API_KEY}" \
  -d "{
    \"prompt\": \"${INSTRUCTIONS}\",
    \"n\": 2,
    \"size\": \"1024x1024\"
  }"

Output:

{
  "created": 1683295449,
  "data": [
    {
      "url": "https://oaidalleapiprodscus.blob.core.windows.net/private/org-zgHnNxrmfCn3EoM43F5XHh2C/user-oHyrpXv0GiOsmYjenJB4DyaV/img-4gxBgW7RB9BxWe5acOebIVe5.png?st=2023-05-05T13%3A04%3A09Z&se=2023-05-05T15%3A04%3A09Z&sp=r&sv=2021-08-06&sr=b&rscd=inline&rsct=image/png&skoid=6aaadede-4fb3-4698-a8f6-684d7786b067&sktid=a48cca56-e6da-484e-a814-9c849652bcb3&skt=2023-05-05T04%3A49%3A44Z&ske=2023-05-06T04%3A49%3A44Z&sks=b&skv=2021-08-06&sig=ZWMPGNIZVzf8YpD4ETHU/KMHcllajhzu%2Bq6gJ95aJ3c%3D"
    },
    {
      "url": "https://oaidalleapiprodscus.blob.core.windows.net/private/org-zgHnNxrmfCn3EoM43F5XHh2C/user-oHyrpXv0GiOsmYjenJB4DyaV/img-R8X7hpnVKw5323PXAdfgdXBK.png?st=2023-05-05T13%3A04%3A09Z&se=2023-05-05T15%3A04%3A09Z&sp=r&sv=2021-08-06&sr=b&rscd=inline&rsct=image/png&skoid=6aaadede-4fb3-4698-a8f6-684d7786b067&sktid=a48cca56-e6da-484e-a814-9c849652bcb3&skt=2023-05-05T04%3A49%3A44Z&ske=2023-05-06T04%3A49%3A44Z&sks=b&skv=2021-08-06&sig=3kI%2BQKEOxGJuLDjc6AJiK5PqPtqVpRTrm7wURRRqm7c%3D"
    }
  ]
}

Train (fine-tune) OpenAI models with custom data

1. Create a jsonl training file and call it mydata.jsonl

{
  "prompt": "Who is Piet Kernbom?'",
  "completion": "Piet Kernbom was a famous baseball player for the Yankees"
}
{
  "prompt": "Where was Piet Kernbom from?",
  "completion": "He is from Suriname."
}
{
  "prompt": "What are some of Piet Kernbom his hobbies",
  "completion": "Magic tricks and cooking."
}

2. Upload the jsonl training file. Run this curl from the same directory the file is located in.

curl https://api.openai.com/v1/files \
  -H "Authorization: Bearer ${OPENAI_API_KEY}" \
  -F purpose="fine-tune" \
  -F file="@mydata.jsonl"

Output:

{
  "object": "file",
  "id": "file-cj5IFUAN43k3BHd1qRd4lcU2",
  "purpose": "fine-tune",
  "filename": "mydata.jsonl",
  "bytes": 291,
  "created_at": 1683300186,
  "status": "uploaded",
  "status_details": null
}

3. Create a "fine-tune" based on the uploaded file

curl https://api.openai.com/v1/fine-tunes \
  -H "Content-Type: application/json" \
  -H "Authorization: Bearer ${OPENAI_API_KEY}" \
  -d '{
    "training_file": "file-cj5IFUAN43k3BHd1qRd4lcU2",
    "model": "davinci"
  }'

Output:

{
  "object": "fine-tune",
  "id": "ft-SyQ7nzBdGJBVWtFD2rpVfQCp",
  "hyperparams": {
    "n_epochs": 4,
    "batch_size": null,
    "prompt_loss_weight": 0.01,
    "learning_rate_multiplier": null
  },
  "organization_id": "org-zgHnNxrmfCn3EoM43F5XHh2C",
  "model": "davinci",
  "training_files": [
    {
      "object": "file",
      "id": "file-cj5IFUAN43k3BHd1qRd4lcU2",
      "purpose": "fine-tune",
      "filename": "mydata.jsonl",
      "bytes": 291,
      "created_at": 1683300186,
      "status": "processed",
      "status_details": null
    }
  ],
  "validation_files": [],
  "result_files": [],
  "created_at": 1683300391,
  "updated_at": 1683300391,
  "status": "pending",
  "fine_tuned_model": null,
  "events": [
    {
      "object": "fine-tune-event",
      "level": "info",
      "message": "Created fine-tune: ft-SyQ7nzBdGJBVWtFD2rpVfQCp",
      "created_at": 1683300391
    }
  ]
}

4. Pull the "fine-tune" endpoint to retrieve the model ID. Once the status is "succeeded" you can curl the new model which identifier you can find under fine_tuned_model.

curl https://api.openai.com/v1/fine-tunes \
  -H "Authorization: Bearer ${OPENAI_API_KEY}"

5. Hit the new model

curl https://api.openai.com/v1/completions   -H 'Content-Type: application/json'   -H "Authorization: Bearer ${OPENAI_API_KEY}"   -d '{
  "model": "davinci:ft-personal-2023-05-05-15-39-01",
  "prompt": "According to the data I trained on, for what team did Piet Kernbom play baseball?",
  "max_tokens": 10,
  "temperature": 0.2
}'

Output:

{
  "id": "cmpl-7Cs7cr7GxXRYkD9YUc9DiXPPx6HJQ",
  "object": "text_completion",
  "created": 1683302336,
  "model": "davinci:ft-personal-2023-05-05-15-39-01",
  "choices": [
    {
      "text": "\n\nThe answer is the New York Yankees.",
      "index": 0,
      "logprobs": null,
      "finish_reason": "length"
    }
  ],
  "usage": {
    "prompt_tokens": 19,
    "completion_tokens": 10,
    "total_tokens": 29
  }
}

Azure

Request:

curl "https://[resource].openai.azure.com/openai/deployments/[deployment]/chat/completions?api-version=[model]" \
-H "Content-Type: application/json" \
-H "api-key: ${AZURE_API_KEY}" \
-d '{
"messages": [{"role": "user", "content": "What is the OpenAI mission?"}],
"max_tokens": 800,
"temperature": 0.7,
"frequency_penalty": 0,
"presence_penalty": 0,
"top_p": 0.95,
"stop": null,
"stream": true
}'

Response:

{
  "id": "chatcmpl-8SFrsgGImGyykR82c2KhdQ40B06rq",
  "object": "chat.completion",
  "created": 1701744872,
  "model": "gpt-4-32k",
  "prompt_filter_results": [
    {
      "prompt_index": 0,
      "content_filter_results": {
        "hate": {
          "filtered": false,
          "severity": "safe"
        },
        "self_harm": {
          "filtered": false,
          "severity": "safe"
        },
        "sexual": {
          "filtered": false,
          "severity": "safe"
        },
        "violence": {
          "filtered": false,
          "severity": "safe"
        }
      }
    }
  ],
  "choices": [
    {
      "index": 0,
      "finish_reason": "stop",
      "message": {
        "role": "assistant",
        "content": "The mission of OpenAI is to ensure that artificial general intelligence (AGI) benefits all of humanity. OpenAI aims to build safe and beneficial AGI directly, but it is also committed to aiding others in achieving this outcome. It focuses on long-term safety, technical leadership, and a cooperative orientation to actively cooperate with other research and policy institutions and create a global community working together to address AGI’s global challenges."
      },
      "content_filter_results": {
        "hate": {
          "filtered": false,
          "severity": "safe"
        },
        "self_harm": {
          "filtered": false,
          "severity": "safe"
        },
        "sexual": {
          "filtered": false,
          "severity": "safe"
        },
        "violence": {
          "filtered": false,
          "severity": "safe"
        }
      }
    }
  ],
  "usage": {
    "prompt_tokens": 14,
    "completion_tokens": 84,
    "total_tokens": 98
  }
}