Add complete model card

README.md

@@ -1,3 +1,64 @@
 ---
 license: llama3.2
+base_model: meta-llama/Llama-3.2-1B-Instruct
+pipeline_tag: text-generation
+library_name: rkllm
+tags:
+- rkllm
+- rockchip
+- rk3576
+- rk3588
+- llama
+- llama-3.2
 ---
+
+# Llama-3.2-1B-Instruct-RKLLM
+
+RKLLM-converted versions of Llama 3.2 1B Instruct for deployment on Rockchip RK3576 and RK3588 NPUs.
+
+`.rkllm` files require the Rockchip RKLLM runtime and cannot be loaded directly with Transformers, llama.cpp, or Ollama.
+
+## Base Model
+
+- Model: Llama 3.2 1B Instruct
+- Author: Meta
+- Original model: [meta-llama/Llama-3.2-1B-Instruct](https://huggingface.co/meta-llama/Llama-3.2-1B-Instruct)
+- Original license: Llama 3.2 Community License
+
+Access to the upstream model is gated. Refer to its model card, license, and acceptable-use policy before using these converted artifacts.
+
+## Available Variants
+
+| Target | File | Quantization |
+| --- | --- | --- |
+| RK3576 | [`Llama-3.2-1B-Instruct_RK3576_w4a16_g128.rkllm`](RK3576/Llama-3.2-1B-Instruct_RK3576_w4a16_g128.rkllm) | W4A16 (g128) |
+| RK3576 | [`Llama-3.2-1B-Instruct_RK3576_w8a8.rkllm`](RK3576/Llama-3.2-1B-Instruct_RK3576_w8a8.rkllm) | W8A8 |
+| RK3588 | [`Llama-3.2-1B-Instruct_RK3588_w8a8.rkllm`](RK3588/Llama-3.2-1B-Instruct_RK3588_w8a8.rkllm) | W8A8 |
+
+The repository also contains `Llama-3.2-1B-Instruct_data_quant.json`, the calibration data used during conversion.
+
+## Usage
+
+Download the model variant that matches your target SoC:
+
+```bash
+hf download HanzoHuang/Llama-3.2-1B-Instruct-RKLLM \
+  RK3576/Llama-3.2-1B-Instruct_RK3576_w4a16_g128.rkllm \
+  --local-dir Llama-3.2-1B-Instruct-RKLLM
+```
+
+Load the `.rkllm` file with a compatible RKLLM runtime and apply the Llama 3.2 Instruct chat template in your application.
+
+## Limitations
+
+- These are hardware-specific converted artifacts, not Transformers checkpoints.
+- Runtime, driver, and toolkit compatibility depends on the Rockchip software stack installed on the device.
+- Conversion may change output quality relative to the upstream floating-point model; validate on your own workload.
+
+## License
+
+Use of these converted artifacts remains subject to the Llama 3.2 Community License and applicable upstream terms.
+
+## Acknowledgements
+
+Thanks to Meta for releasing Llama 3.2 and to Rockchip and RKLLM contributors for the deployment toolchain.