Introducing CodeLlama 70B: A 70 billion-parameter model achieving SOTA performance in code generation.

Welcome to the?latest edition of the AI in 5?newsletter with Clarifai!

Every week, we bring you the models, tools and tips to build production-ready AI!

This week, we bring you: ??

• Cross-modal search
• Brand new models:?Open AI Text Embedding models, CodeLlama 70B
• Databricks Connect UI module
• AI tip of the week: Passing inference parameters to Whisper model.

Cross-modal search ??

Cross-modal?search refers to the ability to search across different modalities. This will allow the user to search for images, videos, and audio using natural language and vice versa.?

The following blog guides you through cross-modal search and helps you learn how to perform text-to-image search with Clarifai using both the API and the portal UI. ??

Read the blog here.

New models in the Clarifai platform ??

• Text-Embedding-3-large:?This is the larger, more powerful, and one of the best-performing embedding models from Open AI, which can create embeddings with up to 3072 dimensions. The model's average score has increased from 31.4% to 54.9% on the Multi-Language Retrieval (MIRACL) benchmark, while on the commonly used benchmark for English tasks (MTEB), the average score has increased from 61.0% to 64.6%.
• Text-Embedding-3-small:?This is a smaller and considerably more efficient embedding model than the previous generation text-embedding-ada-002 model.The pricing for text-embedding-3-small has reduced by 5X compared to text-embedding-ada-002.
• CodeLlama 70B Instruct:?CodeLlama-70b-Instruct?is a SOTA LLM specialized in code synthesis and understanding. The model is part of the Code Llama series and is optimized for processing and generating code based on natural language instructions. The model achieves 67.8% on HumanEval, reaching the initial GPT-4 performance.
• CodeLlama 70B Python: The Code Llama-70b-Python model is more tailored for Python programming, providing high accuracy and efficiency in Python-related tasks, achieving 57.3% in the HumanEval benchmark.

Databricks Connect UI module ??

Clarifai-PySpark?integration helps interact between Databricks and Clarifai for tasks related to uploading client datasets, annotating data, and exporting and storing annotations.

You can use this Databricks Connect module to:

1. Authenticate a Databricks connection and connect with its compute clusters.

2. Export data and annotations from a Clarifai app into Databricks volume and table.

3. Import data from Databricks volume into the Clarifai app and dataset.

4. Update annotation information within the chosen Delta table for the Clarifai app whenever annotations are getting updated.

Try out the module here.

AI tip of the week: ??

Passing inference parameters to Whisper Model.

When accessing the?Whisper?model from the Clarifai community, you can also pass your inference parameters, such as "task",?to decide whether to translate or transcribe an audio file.

The following code shows how you can translate and transcribe a Spanish audio file using Clarifai's Python SDK.?Check out the code here.

Want to learn more from Clarifai? “Subscribe” to make sure you don’t miss the latest news, tutorials, educational materials, and tips. Thanks for reading!