Add complete model card

README.md

@@ -1,3 +1,59 @@
 ---
 license: apache-2.0
+base_model: google/gemma-4-E2B-it
+library_name: rkllm
+tags:
+- rkllm
+- rockchip
+- rk3576
+- rk3588
+- gemma
+- gemma-4
 ---
+
+# gemma-4-E2B-it-RKLLM
+
+RKLLM-converted versions of Gemma 4 E2B IT 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: Gemma 4 E2B IT
+- Author: Google
+- Original model: [google/gemma-4-E2B-it](https://huggingface.co/google/gemma-4-E2B-it)
+- Original license: Apache-2.0
+
+Refer to the upstream model card for the original model's capabilities, limitations, and acceptable-use guidance.
+
+## Available Variants
+
+| Target | File | Quantization |
+| --- | --- | --- |
+| RK3576 | [`gemma-4-E2B-it_RK3576_w4a16_g128.rkllm`](RK3576/gemma-4-E2B-it_RK3576_w4a16_g128.rkllm) | W4A16 (g128) |
+| RK3576 | [`gemma-4-E2B-it_RK3576_w8a8.rkllm`](RK3576/gemma-4-E2B-it_RK3576_w8a8.rkllm) | W8A8 |
+| RK3588 | [`gemma-4-E2B-it_RK3588_w8a8.rkllm`](RK3588/gemma-4-E2B-it_RK3588_w8a8.rkllm) | W8A8 |
+
+The repository also contains `gemma-4-E2B-it_data_quant.json`, the calibration data used during conversion.
+
+## Usage
+
+Download the model variant that matches your target SoC:
+
+```bash
+hf download HanzoHuang/gemma-4-E2B-it-RKLLM \
+  RK3576/gemma-4-E2B-it_RK3576_w4a16_g128.rkllm \
+  --local-dir gemma-4-E2B-it-RKLLM
+```
+
+Load the `.rkllm` file with a compatible RKLLM runtime. Use the correct prompt formatting for the instruction-tuned upstream model.
+
+## 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.
+
+## Acknowledgements
+
+Thanks to Google for releasing Gemma 4 E2B IT and to Rockchip and RKLLM contributors for the deployment toolchain.