bitnet.cpp is the official inference framework for 1-bit LLMs (e.g., BitNet b1.58). It offers a suite of optimized kernels, that support fast and lossless inference of 1.58-bit models on CPU (with NPU and GPU support coming next).

13 hours ago · BitNet b1.58 2B4T is a native 1-bit LLM trained at scale; it only takes up 400MB, compared to other “small models” that can reach up to 4.8 GB. BitNet b1.58 2B4T model performance, purpose ...

2 days ago · BitNet b1.58 2B4T, the first 1-bit model with 2 billion parameters, is developed to run efficiently even on CPUs like Apple’s M2. What makes BitNet special isn’t just its compact size, it’s the way it works. Instead of relying on full-precision weights, BitNet compresses them into just three values: -1, 0, and 1.

2 days ago · Microsoft researchers claim they’ve developed the largest-scale 1-bit AI model, also known as a “bitnet,” to date. Called BitNet b1.58 2B4T, it’s openly available under an MIT license and ...

Official inference framework for 1-bit LLMs. Contribute to microsoft/BitNet development by creating an account on GitHub.

2 days ago · We introduce BitNet b1.58 2B4T, the first open-source, native 1-bit Large Language Model (LLM) at the 2-billion parameter scale. Trained on a corpus of 4 trillion tokens, the model has been rigorously evaluated across benchmarks covering language understanding, mathematical reasoning, coding proficiency, and conversational ability. Our results …

2 days ago · The Microsoft researchers say that BitNet b1.58 2B4T is the first bitnet with 2 billion parameters, “parameters” being largely synonymous with “weights.”

In this work, we introduce BitNet, a scalable and stable 1-bit Transformer architecture designed for large language models. Specifically, we introduce BitLinear as a drop-in replacement of the nn.Linear layer in order to train 1-bit weights from scratch.

Oct 17, 2023 · In this work, we introduce BitNet, a scalable and stable 1-bit Transformer architecture designed for large language models. Specifically, we introduce BitLinear as a drop-in replacement of the this http URL layer in order to train 1-bit weights from scratch.

bitnet.cpp is the official inference framework for 1-bit LLMs (e.g., BitNet b1.58). It offers a suite of optimized kernels, that support fast and lossless inference of 1.58-bit models on CPU (with NPU and GPU support coming next).