Open AI Completion

# This file tells Git which files shouldn't be added to source control

# First, specify the base Docker image.
# You can see the Docker images from Apify at https://hub.docker.com/r/apify/.
# You can also use any other image from Docker Hub.
FROM apify/actor-python:3.9

# Second, copy just requirements.txt into the actor image,
# since it should be the only file that affects "pip install" in the next step,
# in order to speed up the build
COPY requirements.txt ./

# Install the packages specified in requirements.txt,
# Print the installed Python version, pip version
# and all installed packages with their versions for debugging
RUN echo "Python version:" \
 && python --version \
 && echo "Pip version:" \
 && pip --version \
 && echo "Installing dependencies from requirements.txt:" \
 && pip install -r requirements.txt \
 && echo "All installed Python packages:" \
 && pip freeze

# Next, copy the remaining files and directories with the source code.
# Since we do this after installing the dependencies, quick build will be really fast
# for most source file changes.
COPY . ./

# Specify how to launch the source code of your actor.
# By default, the main.py file is run
CMD python3 main.py

{
  "title": "Input schema for the example_python actor.",
  "type": "object",
  "schemaVersion": 1,
  "properties": {
    "api_key": {
      "title": "API key",
      "type": "string",
      "description": "Enter [OpenAI API Key](https://beta.openai.com/docs/developer-quickstart/your-api-keys)",
      "editor": "textfield"
    },
    "prompt": {
      "title": "Prompt",
      "type": "array",
      "description": "Enter OpenAI prompt message",
      "editor": "stringList",
      "minItems": 1
    },
    "engine": {
      "title": "Engine",
      "type": "string",
      "description": "Select <a href='https://beta.openai.com/docs/api-reference/engines'>OpenAI engine</a>",
      "editor": "select",
      "default": "davinci",
      "prefill": "davinci",
      "enum": ["davinci","curie","babbage","ada","davinci-instruct-beta","curie-instruct-beta", "davinci-codex", "cushman-codex", "content-filter-alpha"],
      "enumTitles": ["Base: DaVinci","Base: Curie", "Base: Babbage", "Base: Ada", "Instruct: DaVinci", "Instruct: Curie", "Codex: DaVinci", "Codex: Cushman", "Content filter"]
    },
    "temperature": {
      "title": "Temperature",
      "type": "string",
      "description": "What sampling temperature to use. Higher values means the model will take more risks. Try 0.9 for more creative applications, and 0 (argmax sampling) for ones with a well-defined answer.<br />We generally recommend altering this or top_p but not both.",      
      "editor": "textfield",
      "default": "0"
    },
    "max_tokens": {
      "title": "Max. tokens",
      "type": "integer",
      "description": "The maximum number of tokens to generate in the completion.<br />The token count of your prompt plus max_tokens cannot exceed the model's context length. Most models have a context length of 2048 tokens (except davinci-codex, which supports 4096).",
      "default": 16,
      "maximum": 4096,
      "minimum": 1
    },
    "top_p": {
      "title": "Top p",
      "type": "string",
      "description": "An alternative to sampling with temperature, called nucleus sampling, where the model considers the results of the tokens with top_p probability mass. So 0.1 means only the tokens comprising the top 10% probability mass are considered.<br />We generally recommend altering this or temperature but not both.",
      "editor": "textfield",
      "default": "1"
    },
    "frequency_penalty": {
      "title": "Frequency penalty",
      "type": "string",
      "description": "Number between -2.0 and 2.0. Positive values penalize new tokens based on their existing frequency in the text so far, decreasing the model's likelihood to repeat the same line verbatim.",
      "editor": "textfield",
      "default": "0"
    },
    "presence_penalty": {
      "title": "Presence penalty",
      "type": "string",
      "description": "Number between -2.0 and 2.0. Positive values penalize new tokens based on whether they appear in the text so far, increasing the model's likelihood to talk about new topics.",
      "editor": "textfield",
      "default": "0"
    },
    "stop": {
      "title": "Stop sequence",
      "type": "array",
      "description": "Define one or more sequences that when generated force GPT-3 to stop.",
      "editor": "json",
      "prefill": ["\n"],
      "default": ["\n"]
    }
  },
  "required": ["api_key","prompt"]
}

{
	"name": "openai-completion-actor",
	"version": "0.0",
	"buildTag": "latest",
	"env": null
}

1import json
2import os
3import openai
4
5from apify_client import ApifyClient
6
7
8# Run the main function of the script, if the script is executed directly
9if __name__ == '__main__':
10    # Initialize the main ApifyClient instance
11    client = ApifyClient(os.environ['APIFY_TOKEN'], api_url=os.environ['APIFY_API_BASE_URL'])
12
13    # Get the resource subclient for working with the default key-value store of the actor
14    default_kv_store_client = client.key_value_store(os.environ['APIFY_DEFAULT_KEY_VALUE_STORE_ID'])
15
16    # Get the value of the actor input and print it
17    actor_input = default_kv_store_client.get_record(os.environ['APIFY_INPUT_KEY'])['value']
18    print('Actor input:')
19    print(json.dumps(actor_input, indent=2))
20
21    # Get OpenAI API key (https://beta.openai.com/docs/developer-quickstart/your-api-keys)
22    openai.api_key = actor_input["api_key"]
23
24    responses = []
25    dataset = []
26    # Walk through all the prompts (using the same other parameters)
27    for prompt in actor_input['prompt']:
28        response = openai.Completion.create(
29            engine=actor_input['engine'],
30            prompt=prompt,
31            temperature=float(actor_input['temperature']),
32            max_tokens=actor_input['max_tokens'],
33            top_p=float(actor_input['top_p']),
34            frequency_penalty=float(actor_input['frequency_penalty']),
35            presence_penalty=float(actor_input['presence_penalty']),
36            stop=actor_input['stop']
37        )
38        responses.append(response.choices[0])
39        dataset.append({
40                "prompt": prompt,
41                "response": response.choices[0].text
42            })
43
44    # Set the 'OUTPUT' key-value store record to the same value as the input
45    default_kv_store_client.set_record('OUTPUT', responses)
46
47    # Get the resource subclient for working with the default dataset of the actor
48    default_dataset_client = client.dataset(os.environ['APIFY_DEFAULT_DATASET_ID'])
49
50    # Push responses to the default dataset
51    default_dataset_client.push_items(dataset)

1# Add your dependencies here.
2# See https://pip.pypa.io/en/latest/cli/pip_install/#requirements-file-format
3# for how to format them
4openai