Merge branch 'main' of https://github.com/intel/llm-scaler

Author: winsondu

vllm/Miner-U/README.md

@@ -0,0 +1,55 @@
+Running Miner-U on intel BMG cards
+
+## 1. Start docker and download models
+
+```bash
+docker pull intel/llm-scaler-vllm:0.10.0-b3
+export DOCKER_IMAGE=intel/llm-scaler-vllm:0.10.0-b3
+name="test-mineru"
+docker rm -f $name
+docker run -itd \
+        --net=host \
+        --device=/dev/dri \
+        --privileged \
+        --entrypoint /bin/bash \
+        --name=$name \
+        -v /home/intel/LLM:/llm/models/ \
+        -e http_proxy=your_http_proxy \
+        -e https_proxy=your_http_proxy \
+        -e HF_ENDPOINT=https://hf-mirror.com \
+        -e no_proxy="127.0.0.1,localhost" \
+        --shm-size="16g" \
+        $DOCKER_IMAGE
+
+docker exec -it test-mineru /bin/bash
+mineru-models-download
+# choose huggingface
+# choose pipeline
+```
+
+
+## 2. start miner-u and use
+
+offline running:
+```bash
+mineru -p /llm/MinerU/demo/pdfs/small_ocr.pdf -o ./ --source local --device xpu
+```
+
+start api-serving to use:
+```bash
+export MINERU_DEVICE_MODE="xpu"
+mineru-api --host 0.0.0.0 --port 8008
+
+time curl -X POST "http://127.0.0.1:8008/file_parse" \
+  -F "files=@/llm/MinerU/demo/pdfs/small_ocr.pdf" \
+  -F "output_dir=./output" \
+  -F "lang_list=ch"
+```
+
+start gradio web-ui to use:
+```bash
+export MINERU_DEVICE_MODE="xpu"
+mineru-gradio --server-name 0.0.0.0 --server-port 7860
+```
+
+Refer to [here](https://opendatalab.github.io/MinerU/zh/usage/quick_usage/#_2) for more details.