Accelerators
Partner Solutions
Professional Services
Collections
Sign In
Accelerators
Partner Solutions
Professional Services
Collections
PlugML
by Internal Labs
0
Activity
7
Summary
Summary
Overview
Overview
What is PlugML Activity
PlugML is a package that will host various models and enable us to quickly solve issues with ML.
· SemanticSimilarity: Allows you to compare two strings and see if they are referring to the same thing.
· Stable Diffusion: Generates an image based on a given text (TextToImage) or modifies an image based on a given text (ImageToImage).
Behind the scenes we use the models available on HuggingFace and we will try to bring more in the future.
How to use
· In order for the activity to work, Python needs to be installed. Latest knowing version of Python for this activity to work with is 3.10.9.
A full guide on how to use this set of activities can be found here.
Semantic Similarity
GenerateEmbedding fields:
1. BatchSize: Splits the given list in smaller batch size to mitigate the chance of running out of memory. Default 64.
2. Data: Array of sentences. Embedding will be created from this.
3. EmbeddingFilePath: The location where the embed file will be created. Example "C:\Users\USERNAME\Downloads". Default is the root of the project.
4. EmbeddingFileName: The name of the embed file that will be created.
5. PythonInstalledPath: Home path where Python is installed in the local machine. Example "C:\Users\USERNAME\AppData\Local\Programs\Python\Python310".
6. PythonLibrarypath: Library path (Linux or version> 3.9); InArgument<Strings>. For Linux is the path to Python libpython*.so library including library name. For Windows (Version>3.9) path to python**.dll including library name (usually is in Python Home path). For Windows (Version <=3,9) leave empty. Example "C:\Users\USERNAME\AppData\Local\Programs\Python\Python310\python310.dll".
Query fields:
1. Data: Array of sentences. The query will be checked against this.
2. QuertSentence: Query to check against sentences.
3. Threshold: Threshold for the ML Model for the cutoff point.
4. AllResults: Will return all matched results above the threshold, not just one.
5. EmbeddingFilePath: The location where the embed file will be created. Example "C:\Users\USERNAME\Downloads". Default is the root of the project.
6. EmbendingFileName: The name of the embed file that will be created.
7. Results: ML response with the best similarity match.
8. PythonInstalledPath: Home path where Python is installed in the local machine. Example "C:\Users\USERNAME\AppData\Local\Programs\Python\Python310".
9. PythonLibrarypath: Library path (Linux or version> 3.9); InArgument<Strings> For Linux is the path to Python libpython*.so library including library name. For Windows (Version>3.9) path to python**.dll including library name (usually is in Python Home path). For Windows (Version <=3,9) leave empty. Example "C:\Users\USERNAME\AppData\Local\Programs\Python\Python310\python310.dll".
StableDiffusion
Note: You will need to create a token on huggingface to be able to use these activities.
TextToImage fields:
1. Prompt: Text description of the image that will be generated.
2. ImageGuidanceScale: Used to increase the adherence to the prompt that guides the image generation.
3. FilePath: The path of the image to be generated. Example "C:\Users\USERNAME\Downloads".
4. Token: Hugging Face token, which is necessary for accessing the stable diffusion models.
5. Timeout: Specifies the amount of time (as Seconds) for the activity to run, before an exception is thrown and the activity is aborted.
6. PythonInstalledPath: Home path where Python is installed in the local machine. Example "C:\Users\USERNAME\AppData\Local\Programs\Python\Python310".
7. PythonLibrarypath: Library path (Linux or version> 3.9); InArgument<Strings> For Linux is the path to Python libpython*.so library including library name. For Windows (Version>3.9) path to python**.dll including library name (usually is in Python Home path). For Windows (Version <=3,9) leave empty. Example "C:\Users\USERNAME\AppData\Local\Programs\Python\Python310\python310.dll".
ImageToImage fields:
1. ImageGuidanceScale: Used to increase the adherence to the prompt that guides the image generation.
2. Prompt: Text description of the image that will be generated.
3. FilePath: The path of the image to be generated.
Example "C:\Users\USERNAME\Downloads".
4. ImageStrength: Value between 0.0 and 1.0, that controls the amount of noise that is added to the input image.
5. Timeout: Specifies the amount of time (in Seconds) for the activty to run, before an exception is thrown and the activity is aborted.
6. Token: Hugging Face token, which is necessary for accessing the stable diffusion models.
7. ImageURL: URL of the input image, which will be modified based on the text prompt.
8. PythonInstalledPath: Home path where Python is installed in the local machine.
Example "C:\Users\USERNAME\AppData\Local\Programs\Python\Python310".
9. PythonLibrarypath: Library path (Linux or version> 3.9); InArgument<Strings>
Example "C:\***\Python\Python310\python310.dll".
Features
Features
Plug ML has 2 main components at the moment:
1. Semantic Similarity - A module designed to help with fuzzy matching (especially useful for addresses).
2. Stable Diffusion - Image to Image or Text to Image generation of AI Images.
Additional Information
Additional Information
Dependencies
Code Language
Visual Basic
Runtime
Windows (.Net 5.0 or higher)
Technical
Version
1.0.0Updated
Jan 26, 2023Works with
Studio: 22.10+
Certification
Silver Certified
Tags
fuzzy search
stable diffusion
ml
semantic
Application
UiPath Studio
Support
UiPath Community Support
Resources
Similar Listings
