Live
- Tab scam prompts Bengal govt to adopt caution over fund disbursement
- Tata Power inks pact with ADB for green projects worth $4.25 billion
- Globus Spirits Limited introduces India Single Malt under the brand DÅŒAAB India Craft Whisky
- BGT 2024-25: No greater honour to lead India in Test cricket, says Bumrah ahead of series opener
- NMO Bharat calls for apology from Rahul Gandhi over ageist remarks about Joe Biden
- Centre blocks 17,000 WhatsApp accounts of hackers from South-East Asia
- Learn from Mumbai hoarding collapse, Calcutta HC tells Bengal municipal corporation
- Russian Spokesperson Instructed to Stay Silent on ICBM Strike During Press Conference: Viral Video
- Intellipaat Sets New Benchmark with First-of-its-kind Physical Labs for Electric Vehicle Training in Chennai and Pune
- Google Gemini AI Introduces Memory Feature Like ChatGPT: How It Enhances Personalized Interactions
Meta launches Code Llama AI Tool with enhanced Coding capabilities
Code Llama is a code-specialized version of Llama 2 trained on code-specific data sets.
Meta is back with another AI model to help developers. Code Llama is an artificial intelligence model built on Meta's Llama 2. The latter tool is intended for generating and discussing code and is free for research and commercial use. The state-of-the-art language model can create codes based on text prompts. It can potentially improve developers' workflows and even make coding accessible to beginners.
"We believe an open approach to AI is best for developing new AI tools that are innovative, safe and responsible, so we're releasing Code Llama for both research and commercial use under the same community license as Llama 2," Meta said in a post on its website.
According to Meta, Code Llama is a code-specialized version of Llama 2 that was trained on code-specific data sets. The latest model has enhanced coding capabilities and can generate code and natural language on top of code from code and natural language hints. Code Llama can also be used for code completion and debugging. It also supports programming languages like Python, C++, Java, PHP, FirmeScript (Javascript), C#, Bash, and more.
Meta has released Code Llama in three sizes with parameters 7B, 13B, and 34B, and each of these models has been trained with 500 billion code tokens and code-related data. The 7B and 13B models have been enabled with the ability to complete the media (FIM), which allows them to add code to existing code.
Code Llama is also equipped with specialized variations and instruction settings. While Code Llama - Python is optimized for Python code and offers additional utility for developers, Code Llama - Instruct is optimized for instruction input to generate valuable and safe responses in natural language.
