Meta Platforms: Artificial Intelligence(AI) News: Difference between revisions

Meta Platforms: Artificial Intelligence(AI) News (view source)

Revision as of 13:55, 24 August 2023

5 bytes removed , 24 August 2023

→‎AI Developments

VisualWikitext

Aron

Administrators

6,294

edits

@@ Line 4: / Line 4: @@
 == AI Developments ==
-August 24, 2023: Meta said introduces an AI model that will assist in writing computer code. The model known as Code Llama will use text prompts to generate codes. It can also be used to debug and complete written codes. Code Llama is built using Llama 2 language model. The model will be available in different configurations and will support languages such as Python, Java and C++. However, the model is not recommended for general text tasks. Meta said that their test on Code Llama found that it's on par with ChatGPT and better than other open-sourced LLMs including Llama 2. Like its other LLMs, Code Llama is free for research and commercial use<ref>https://ai.meta.com/blog/code-llama-large-language-model-coding/</ref>.
+August 24, 2023: Meta introduces an AI model that will assist in writing computer code. The model known as Code Llama will use text prompts to generate codes. It can also be used to debug and complete written codes. Code Llama is built using Llama 2 language model. The model will be available in different configurations and will support languages such as Python, Java and C++. However, the model is not recommended for general text tasks. Meta said that their test on Code Llama found that it's on par with ChatGPT and better than other open-sourced LLMs including Llama 2. Like its other LLMs, Code Llama is free for research and commercial use<ref>https://ai.meta.com/blog/code-llama-large-language-model-coding/</ref>.
 August 22, 2023: Meta releases SeamlessM4T AI model that enables translation between speech and text for up to 100 languages. It also enables speech-to-speech translation for up to 35 languages. The company will make the model available for the public for non-commercial purposes.  Meta said the training data for the model came from 4 million hours of "raw audio originating from a publicly available repository of crawled web data." The datasets came from Wikipedia and similar websites<ref>https://finance.yahoo.com/news/meta-releases-ai-model-translating-133311959.html</ref>. Over time, Meta plans to integrate these AI advances into Facebook, Instagram, WhatsApp, Mesenger, and Threads<ref>https://www.investmentwiki.org/w/index.php?title=Meta_Platforms:_Artificial_Intelligence(AI)&action=edit&section=1</ref>.