Meta Publishes Complete Llama 3.1 405B Weights Under Open License
Meta released the full 405 billion parameter weights for Llama 3.1 with a commercial use license. Developers can download the model files directly and run inference on their own hardware. The release includes the tokenizer and configuration files needed for local fine tuning.
Teams move from paying per token to owning their inference costs and data boundaries. Local deployment removes the requirement to send proprietary data to third party servers. Organizations gain the ability to customize model behavior through continued pre training on domain specific corpora.
Hugging Face hosts the weights at huggingface.co/meta-llama/Llama-3.1-405B and reports over 500,000 downloads in the first week. Several university research groups have already published fine tuned variants for medical question answering that outperform the base model on their benchmarks.
Step 1: Create a Hugging Face account and request access at huggingface.co/meta-llama/Llama-3.1-405B. Step 2: Use the transformers library to load the model with from_pretrained('meta-llama/Llama-3.1-405B') on a machine with at least 8 A100 GPUs. Step 3: Run inference with your own prompts and save the resulting model checkpoint after any fine tuning passes.