openai model

Former-commit-id: 20b1020a2e

apps/paper_swarm.py

@@ -15,6 +15,7 @@ import requests
 from bs4 import BeautifulSoup
 import os
 
+
 class Paper:
     def __init__(self, title, date, authors, abstract):
         self.title = title
@@ -22,33 +23,37 @@ class Paper:
         self.authors = authors
         self.abstract = abstract
 
+
 class Scraper:
     def __init__(self, url):
         self.url = url
 
     def get_paper_links(self):
         response = requests.get(self.url)
-        soup = BeautifulSoup(response.text, 'html.parser')
+        soup = BeautifulSoup(response.text, "html.parser")
-        links = [a['href'] for a in soup.find_all('a', href=True) if '/papers/' in a['href']]
+        links = [
+            a["href"] for a in soup.find_all("a", href=True) if "/papers/" in a["href"]
+        ]
         return links
 
     def get_paper_details(self, link):
         response = requests.get(self.url + link)
-        soup = BeautifulSoup(response.text, 'html.parser')
+        soup = BeautifulSoup(response.text, "html.parser")
-        title = soup.find('h1').text
+        title = soup.find("h1").text
-        date_tag = soup.find('time')
+        date_tag = soup.find("time")
-        date = date_tag.text if date_tag else 'Unknown'
+        date = date_tag.text if date_tag else "Unknown"
-        authors = [author.text for author in soup.find_all('span', class_='author')]
+        authors = [author.text for author in soup.find_all("span", class_="author")]
-        abstract_tag = soup.find('div', class_='abstract')
+        abstract_tag = soup.find("div", class_="abstract")
-        abstract = abstract_tag.text if abstract_tag else 'Abstract not found'
+        abstract = abstract_tag.text if abstract_tag else "Abstract not found"
         return Paper(title, date, authors, abstract)
 
+
 class FileWriter:
     def __init__(self, directory):
         self.directory = directory
 
     def write_paper(self, paper):
-        with open(os.path.join(self.directory, paper.title + '.txt'), 'w') as f:
+        with open(os.path.join(self.directory, paper.title + ".txt"), "w") as f:
             f.write(f"h1: {paper.title}\n")
             f.write(f"Published on {paper.date}\n")
             f.write("Authors:\n")
@@ -57,8 +62,9 @@ class FileWriter:
             f.write("Abstract\n")
             f.write(paper.abstract)
 
-scraper = Scraper('https://huggingface.co/papers')
+
-file_writer = FileWriter('images')
+scraper = Scraper("https://huggingface.co/papers")
+file_writer = FileWriter("images")
 
 links = scraper.get_paper_links()
 for link in links:

docs/swarms/models/openai.md

@@ -0,0 +1,196 @@
+# `BaseOpenAI` and `OpenAI` Documentation
+
+## Table of Contents
+
+1. [Overview](#overview)
+2. [Class Architecture](#class-architecture)
+3. [Purpose](#purpose)
+4. [Class Attributes](#class-attributes)
+5. [Methods](#methods)
+   - [Construction](#construction)
+   - [Configuration](#configuration)
+   - [Tokenization](#tokenization)
+   - [Generation](#generation)
+   - [Asynchronous Generation](#asynchronous-generation)
+6. [Usage Examples](#usage-examples)
+   - [Creating an OpenAI Object](#creating-an-openai-object)
+   - [Generating Text](#generating-text)
+   - [Advanced Configuration](#advanced-configuration)
+
+---
+
+## 1. Overview <a name="overview"></a>
+
+The `BaseOpenAI` and `OpenAI` classes are part of the LangChain library, designed to interact with OpenAI's large language models (LLMs). These classes provide a seamless interface for utilizing OpenAI's API to generate natural language text.
+
+## 2. Class Architecture <a name="class-architecture"></a>
+
+Both `BaseOpenAI` and `OpenAI` classes inherit from `BaseLLM`, demonstrating an inheritance-based architecture. This architecture allows for easy extensibility and customization while adhering to the principles of object-oriented programming.
+
+## 3. Purpose <a name="purpose"></a>
+
+The purpose of these classes is to simplify the interaction with OpenAI's LLMs. They encapsulate API calls, handle tokenization, and provide a high-level interface for generating text. By instantiating an object of the `OpenAI` class, developers can quickly leverage the power of OpenAI's models to generate text for various applications, such as chatbots, content generation, and more.
+
+## 4. Class Attributes <a name="class-attributes"></a>
+
+Here are the key attributes and their descriptions for the `BaseOpenAI` and `OpenAI` classes:
+
+| Attribute                 | Description |
+|---------------------------|-------------|
+| `lc_secrets`              | A dictionary of secrets required for LangChain, including the OpenAI API key. |
+| `lc_attributes`           | A dictionary of attributes relevant to LangChain. |
+| `is_lc_serializable()`    | A method indicating if the class is serializable for LangChain. |
+| `model_name`              | The name of the language model to use. |
+| `temperature`             | The sampling temperature for text generation. |
+| `max_tokens`              | The maximum number of tokens to generate in a completion. |
+| `top_p`                   | The total probability mass of tokens to consider at each step. |
+| `frequency_penalty`       | Penalizes repeated tokens according to frequency. |
+| `presence_penalty`        | Penalizes repeated tokens. |
+| `n`                       | How many completions to generate for each prompt. |
+| `best_of`                 | Generates `best_of` completions server-side and returns the "best." |
+| `model_kwargs`            | Holds any model parameters valid for `create` calls not explicitly specified. |
+| `openai_api_key`          | The OpenAI API key used for authentication. |
+| `openai_api_base`         | The base URL for the OpenAI API. |
+| `openai_organization`     | The OpenAI organization name, if applicable. |
+| `openai_proxy`            | An explicit proxy URL for OpenAI requests. |
+| `batch_size`              | The batch size to use when passing multiple documents for generation. |
+| `request_timeout`         | The timeout for requests to the OpenAI completion API. |
+| `logit_bias`              | Adjustment to the probability of specific tokens being generated. |
+| `max_retries`             | The maximum number of retries to make when generating. |
+| `streaming`               | Whether to stream the results or not. |
+| `allowed_special`         | A set of special tokens that are allowed. |
+| `disallowed_special`      | A collection of special tokens that are not allowed. |
+| `tiktoken_model_name`     | The model name to pass to `tiktoken` for token counting. |
+
+## 5. Methods <a name="methods"></a>
+
+### 5.1 Construction <a name="construction"></a>
+
+#### 5.1.1 `__new__(cls, **data: Any) -> Union[OpenAIChat, BaseOpenAI]`
+- Description: Initializes the OpenAI object.
+- Arguments:
+  - `cls` (class): The class instance.
+  - `data` (dict): Additional data for initialization.
+- Returns:
+  - Union[OpenAIChat, BaseOpenAI]: An instance of the OpenAI class.
+
+### 5.2 Configuration <a name="configuration"></a>
+
+#### 5.2.1 `build_extra(cls, values: Dict[str, Any]) -> Dict[str, Any]`
+- Description: Builds extra kwargs from additional params passed in.
+- Arguments:
+  - `cls` (class): The class instance.
+  - `values` (dict): Values and parameters to build extra kwargs.
+- Returns:
+  - Dict[str, Any]: A dictionary of built extra kwargs.
+
+#### 5.2.2 `validate_environment(cls, values: Dict) -> Dict`
+- Description: Validates that the API key and python package exist in the environment.
+- Arguments:
+  - `values` (dict): The class values and parameters.
+- Returns:
+  - Dict: A dictionary of validated values.
+
+### 5.3 Tokenization <a name="tokenization"></a>
+
+#### 5.3.1 `get_sub_prompts(self, params: Dict[str, Any], prompts: List[str], stop: Optional[List[str]] = None) -> List[List[str]]`
+- Description: Gets sub-prompts for LLM call.
+- Arguments:
+  - `params` (dict): Parameters for LLM call.
+  - `prompts` (list): List of prompts.
+  - `stop` (list, optional): List of stop words.
+- Returns:
+  - List[List[str]]: List of sub-prompts.
+
+#### 5.3.2 `get_token_ids(self, text: str) -> List[int]`
+- Description: Gets token IDs using the `tiktoken` package.
+- Arguments:
+  - `text` (str): The text for which to calculate token IDs.
+- Returns:
+  - List[int]: A list of token IDs.
+
+#### 5.3.3 `modelname_to_contextsize(modelname: str) -> int`
+- Description: Calculates the maximum number of tokens possible to generate for a model.
+- Arguments:
+  - `modelname` (str): The model name to determine the context size for.
+- Returns:
+  - int: The maximum context size.
+
+#### 5.3.4 `max_tokens_for_prompt(self, prompt: str) -> int`
+- Description: Calculates the maximum number of tokens possible to generate for a prompt.
+- Arguments:
+  - `prompt` (str): The prompt for which to
+
+ determine the maximum token limit.
+- Returns:
+  - int: The maximum token limit.
+
+### 5.4 Generation <a name="generation"></a>
+
+#### 5.4.1 `generate(self, text: Union[str, List[str]], **kwargs) -> Union[str, List[str]]`
+- Description: Generates text using the OpenAI API.
+- Arguments:
+  - `text` (str or list): The input text or list of inputs.
+  - `**kwargs` (dict): Additional parameters for the generation process.
+- Returns:
+  - Union[str, List[str]]: The generated text or list of generated texts.
+
+### 5.5 Asynchronous Generation <a name="asynchronous-generation"></a>
+
+#### 5.5.1 `generate_async(self, text: Union[str, List[str]], **kwargs) -> Union[str, List[str]]`
+- Description: Generates text asynchronously using the OpenAI API.
+- Arguments:
+  - `text` (str or list): The input text or list of inputs.
+  - `**kwargs` (dict): Additional parameters for the asynchronous generation process.
+- Returns:
+  - Union[str, List[str]]: The generated text or list of generated texts.
+
+## 6. Usage Examples <a name="usage-examples"></a>
+
+### 6.1 Creating an OpenAI Object <a name="creating-an-openai-object"></a>
+
+```python
+# Import the OpenAI class
+from swarms.models import OpenAI
+
+# Set your OpenAI API key
+api_key = "YOUR_API_KEY"
+
+# Create an OpenAI object
+openai = OpenAI(api_key)
+```
+
+### 6.2 Generating Text <a name="generating-text"></a>
+
+```python
+# Generate text from a single prompt
+prompt = "Translate the following English text to French: 'Hello, how are you?'"
+generated_text = openai.generate(prompt, max_tokens=50)
+
+# Generate text from multiple prompts
+prompts = ["Translate this: 'Good morning' to Spanish.", "Summarize the following article:", article_text]
+generated_texts = openai.generate(prompts, max_tokens=100)
+
+# Generate text asynchronously
+async_prompt = "Translate 'Thank you' into German."
+async_result = openai.generate_async(async_prompt, max_tokens=30)
+
+# Access the result of an asynchronous generation
+async_result_text = async_result.get()
+```
+
+### 6.3 Advanced Configuration <a name="advanced-configuration"></a>
+
+```python
+# Configure generation with advanced options
+custom_options = {
+    "temperature": 0.7,
+    "max_tokens": 100,
+    "top_p": 0.9,
+    "frequency_penalty": 0.2,
+    "presence_penalty": 0.4
+}
+generated_text = openai.generate(prompt, **custom_options)
+```
+
+This documentation provides a comprehensive understanding of the `BaseOpenAI` and `OpenAI` classes, their attributes, methods, and usage examples. Developers can utilize these classes to interact with OpenAI's language models efficiently, enabling various natural language generation tasks.

mkdocs.yml

@@ -92,6 +92,7 @@ nav:
     - Overview: "swarms/models/index.md"
     - HuggingFaceLLM: "swarms/models/hf.md"
     - Anthropic: "swarms/models/anthropic.md"
+    - OpenAI: "swarms/modeks/openai.md"
   - swarms.structs:
     - Overview: "swarms/structs/overview.md"
     - Workflow: "swarms/structs/workflow.md"

pyproject.toml

@@ -50,7 +50,9 @@ open-interpreter = "*"
 tabulate = "*"
 termcolor = "*"
 black = "*"
-dalle = "*"
+dalle3 = "*"
+torchvision = "*"
+
 
 [tool.poetry.dev-dependencies]
 first_dependency = {git = "https://github.com/IDEA-Research/GroundingDINO.git"}

requirements.txt

@@ -52,6 +52,7 @@ webdataset
 yapf
 autopep8
 dalle3
+torchvision
 
 
 mkdocs

swarms/agents/__init__.py

@@ -9,4 +9,3 @@ from swarms.agents.stream_response import stream
 from swarms.agents.base import AbstractAgent
 from swarms.agents.registry import Registry
 from swarms.agents.idea_to_image_agent import Idea2Image
-

swarms/agents/idea_to_image_agent.py

@@ -108,4 +108,3 @@ class Idea2Image:
 
         # Download the images to your specified folder
         self.dalle.download(urls, self.output_folder)
-

swarms/models/whisperx.py

@@ -0,0 +1,3 @@
+"""An ultra fast speech to text model."""
+
+

tests/agents/idea_to_image.py

@@ -12,6 +12,7 @@ TEST_OUTPUT_FOLDER = "test_images/"
 OPENAI_API_KEY = openai_key
 DALLE_COOKIE = dalle_cookie
 
+
 @pytest.fixture(scope="module")
 def idea2image_instance():
     # Create an instance of the Idea2Image class
@@ -26,15 +27,18 @@ def idea2image_instance():
     if os.path.exists(TEST_OUTPUT_FOLDER):
         shutil.rmtree(TEST_OUTPUT_FOLDER)
 
+
 def test_idea2image_instance(idea2image_instance):
     # Check if the instance is created successfully
     assert isinstance(idea2image_instance, Idea2Image)
 
+
 def test_llm_prompt(idea2image_instance):
     # Test the llm_prompt method
     prompt = idea2image_instance.llm_prompt()
     assert isinstance(prompt, str)
 
+
 def test_generate_image(idea2image_instance):
     # Test the generate_image method
     idea2image_instance.generate_image()
@@ -44,6 +48,7 @@ def test_generate_image(idea2image_instance):
     files = os.listdir(TEST_OUTPUT_FOLDER)
     assert len(files) > 0
 
+
 def test_invalid_openai_api_key():
     # Test with an invalid OpenAI API key
     with pytest.raises(Exception) as exc_info:
@@ -55,5 +60,6 @@ def test_invalid_openai_api_key():
         )
     assert "Failed to initialize OpenAIChat" in str(exc_info.value)
 
+
 if __name__ == "__main__":
     pytest.main()