🦙 Meta Releases the Largest Open-Source Language Model Llama 3.1-405B

Meta has stunned the AI world by introducing its newest and most powerful open-source language model – Llama 3.1-405B.

Key Facts

⭐️ Parameters: Llama 3.1-405B is the largest model in the Llama series, boasting 405 billion parameters. This latest release also introduced upgraded 8B and 70B parameter model versions.

The number of parameters affects the model’s ability to reason, understand context, and generate diverse, accurate, and creative content. More parameters require more significant computing resources.

📊 Benchmarks: The flagship model is competitive with leading foundation models across various tasks, including GPT-4o and Claude 3.5 Sonnet. See the comparison in language understanding (MMLU), codding (HumanEval), and math (GSM8K and MATH) 🔼

🖼 Multimodality: Llama 3.1 can recognize and generate both text and images. The model has already been integrated into the beta version of WhatsApp for Android.

🔒 Open Source: Llama 3.1 allows developers and researchers to use it in their projects. This makes it accessible to more users, including universities and small companies.

🖥 Context Window: The model uses a new tokenizer that expands the vocabulary from 32K to 128K tokens, improving language processing and allowing more efficient work with text by remembering more context.

➡️ Try Llama 3.1-405B here.

📱 Mark Zuckerberg has already given a video interview.

In the previous series:

❓ Open vs. Closed Source

🦙 Llama 3 Is Out. It's The Largest Open-Source LLM

#news #llama @hiaimediaen