New Delhi, January 30: Meta AI has released a new open-source AI coding assistant, 'Code Llama 70B' for the developers and researchers. The company announced on X its three variants, including CodeLlama - 70B, Code Llama - 70B - Python and Code Llama 70B-Instruct. In its official post on X, the Mark Zuckerberg-owned 'AI at Meta' said, "Today we're releasing Code Llama 70B: a new, more performant version of our LLM for code generation — available under the same license as previous Code Llama models."
According to the reports, the new Code Llama 70B scored 53% accuracy on the HumanEval benchmark. The reports said the scores are better than the ChatGPT-3.5, which stood at 48.1%. According to the reports, the Code Llama 70B score is close to the ChatGPT-4 67%. However, it does not come close to the GPT-4 Turbo, which is a more advanced model. Apple’s Upcoming iOS 18 Software Update Expected To Be Announced at WWDC Event With ‘Biggest’ Update in the iPhone History.
AI at Meta Releases 'Code Llama 70B', Check Announcement on X:
Today we’re releasing Code Llama 70B: a new, more performant version of our LLM for code generation — available under the same license as previous Code Llama models.
Download the models ➡️ https://t.co/fa7Su5XWDC
• CodeLlama-70B-Instruct pic.twitter.com/iZc8fapYEZ
— AI at Meta (@AIatMeta) January 29, 2024
CEO Mark Zuckerberg Announcement on Code Llama 70B:
Meta CEO Mark Zuckerberg announced from his official Facebook account, "We're open sourcing a new and improved Code Llama, including a larger 70B parameter model." The CEO further mentioned that the writing and editing code emerged as essential uses of AI models, and he further stated that the coding ability has been important for models to process information in other domains more logically and rigorously. OpenAI Notified by Italian Regulator About Breaching European Data Privacy Law, Giving the ChatGPT Developer 30 Days To Respond to the Allegations.
As per the report by Indian Express, the latest Gen AI model can write C++, PHP, Java, and Python. The report said these models had been trained using "500 billion tokens" of code and other relevant data. Further, the company also expanded the context window with up to 1,00,000 tokens or likely 75,000 tokens. The report highlighted that the Code Llama is based on the "Llama" AI model, a 175-billion-parameter general-purpose model. The model is accessible to users on platforms like PyTorch and Hugging Face. Mark Zuckerberg also posted on Facebook, "I'm proud of the progress here, and looking forward to including these advances in Llama 3 and future models as well."
(The above story first appeared on LatestLY on Jan 30, 2024 05:02 PM IST. For more news and updates on politics, world, sports, entertainment and lifestyle, log on to our website latestly.com).