Clarifai 10.10: Compute Orchestration [Private-Preview]

This blog post focuses on new features and improvements. For a comprehensive list, including bug fixes, please see the release notes.

Compute Orchestration [Private Preview]

Clarifai’s new Compute Orchestration system provides you with an efficient way to deploy inference workloads on any compute, in any bare metal, or cloud environment. This feature is currently in Private Preview. To request access, please contact us or sign up here.

New Released Models​

• Added Pixtral 12B to the Platform, a cutting-edge multimodal language model from Mistral AI designed to effectively process both natural language and visual inputs including reasoning with charts, figures, and natural scenes. The model also offers a long context window of 128K tokens, allowing it to manage multiple images and substantial amounts of text efficiently.
  

• Added Granite-3.0-8B-Instruct, a versatile, enterprise-ready language model from IBM optimized for multilingual understanding, coding, and instruction-following across diverse tasks and constrained environments.
• Added Granite-3.0-2B-Instruct SLM, a lightweight, multilingual, enterprise-ready language model optimized for instruction-following and code understanding.

Control Center: Added Costs & Budget tab

In the recent release, we have introduced the new Clarifai Control Center, the unified dashboard, a single pane of glass to monitor everything happening within your account on the platform. It serves as the sole source of truth for various information dimensions, enabling you to make informed decisions based on data from multiple sources. 

We have now added a new tab to display users financial data. It displays the costs of your billed operations, making it easier to track and manage expenses on the platform. Read more about Control Center here.

Released text-based, ready-to-use data pre-processing pipelines

The Clarifai Data Utils library provides a variety of tools for handling multimedia data. With the Data Utils library and Python SDK, you can load text files (PDF, DOC, etc.), transform, chunk, and upload them directly to the Clarifai Platform.

The new Data Ingestion Pipeline provides easy-to-use pipelines to load data from files and ingest it into the Clarifai platform. These new pipelines streamline data ingestion into the platform.

Check out this example repository to learn how to use the pre-built foundational pipelines and to create custom pipelines for file partitioning, chunk cleaning, and metadata extraction.

Important changes to the use of PATs and API keys

• Previously, you could use API keys to access any model, concept, or workflow owned by the app scoped to the API key, as well as those owned by the user clarifai in the application main. Now, accessing models or workflows owned by clarifai in the application main can only be done with a PAT tied to your account.
• This change will be rolled out this month in November. Learn more about it here.

Released a UI module for evaluating OCR workflows

The OCR (Optical Character Recognition) Evaluation Module is a powerful tool designed to assess the performance of OCR workflows in Clarifai. It enables users to measure the accuracy and effectiveness of their OCR workflows by comparing model outputs against known ground truth data.

Key features of the evaluation module include:

• Ground truth data management
• Comprehensive OCR performance metrics
• Multiple evaluation methods
• Detailed performance reports
• Support for various OCR workflows

Learn more about the Module here.

Additional changes

• Labeling Tasks: Enabled all assigned reviewers to review each input before the labeling task is marked as complete and also Enabled collaborators to set task priority levels (High, Medium, Low) for each task.
• Added granular scopes for organization’s team members: Users can now assign granular access levels for applications within an organization. When adding team members to an application, you can define specific scopes to limit their access, rather than granting full access to the app. 
• Made various improvements to the Smart Text Search feature: These improvements include replacing the deprecated Language-Understanding workflow with a new text workflow, preventing billing actions when a base workflow is forcefully reindexed, and additional improvements for optimized performance and functionality.

Ready to start building?