This Hugging Face repository provides GGUF-format quantized weights for Unsloth's Gemma 3 4B instruction-tuned model. Developers can download variants for local inference and use the model with compatible runtimes and hardware.
In the Quantised & converted weights space, Gemma 3 4b It takes a focused approach. It focuses on running and fine-tuning an instruction-tuned Gemma 3 model locally with hardware-compatible quantized weights. It is built as an open-source project for developers and machine-learning practitioners. The project is open source (Apache-2.0). Gemma 3 4b It is available on the web, the command line, and API, and it can be self-hosted.
It is developed by Unsloth, and it first shipped in 2023. The project is developed in the open on GitHub with 75.5k stars and 2.6k commits in the last 90 days. The category is crowded — PulseGate's index counts 25 comparable projects. Key capabilities include GGUF Weights, Multiple Quantizations, and BF16 Variant.
Summary written by a language model from the project’s public pages.
What PulseGate has recorded for this listing
Closest matches by what these projects do