Built with Llama. Unmodified mirror of Meta's Llama 3.1 8B base model. No fine-tuning, no quantization, no change to the weights.

This is the base model, not Instruct. It ships no chat template. Use it for completion or as a fine-tuning starting point; for chat, use meta-llama/Llama-3.1-8B-Instruct.

Apple silicon conversions: 4-bit MLX, 8-bit MLX.

Full model card, benchmarks, training details and responsible-use guidance: meta-llama/Llama-3.1-8B.

Llama 3.1 is licensed under the Llama 3.1 Community License, Copyright (c) Meta Platforms, Inc. All Rights Reserved. The full agreement is in LICENSE and reproduced in the gate above. The acceptable use policy is in USE_POLICY.md.