diff --git a/docs/usage/kunlunxin_xpu_deployment.md b/docs/usage/kunlunxin_xpu_deployment.md
index 1699a09c4..b5a938fb6 100644
--- a/docs/usage/kunlunxin_xpu_deployment.md
+++ b/docs/usage/kunlunxin_xpu_deployment.md
@@ -6,21 +6,22 @@
|ERNIE-4.5-300B-A47B|32K|WINT8|8|export XPU_VISIBLE_DEVICES="0,1,2,3,4,5,6,7"
python -m fastdeploy.entrypoints.openai.api_server \
--model PaddlePaddle/ERNIE-4.5-300B-A47B-Paddle \
--port 8188 \
--tensor-parallel-size 8 \
--max-model-len 32768 \
--max-num-seqs 64 \
--quantization "wint8" \
--gpu-memory-utilization 0.9 \
--load-choices "default"|2.3.0|
|ERNIE-4.5-300B-A47B|32K|WINT4|4 (Recommended)|export XPU_VISIBLE_DEVICES="0,1,2,3" or "4,5,6,7"
python -m fastdeploy.entrypoints.openai.api_server \
--model PaddlePaddle/ERNIE-4.5-300B-A47B-Paddle \
--port 8188 \
--tensor-parallel-size 4 \
--max-model-len 32768 \
--max-num-seqs 64 \
--quantization "wint4" \
--gpu-memory-utilization 0.9 \
--load-choices "default"|2.3.0|
|ERNIE-4.5-300B-A47B|32K|WINT4|8|export XPU_VISIBLE_DEVICES="0,1,2,3,4,5,6,7"
python -m fastdeploy.entrypoints.openai.api_server \
--model PaddlePaddle/ERNIE-4.5-300B-A47B-Paddle \
--port 8188 \
--tensor-parallel-size 8 \
--max-model-len 32768 \
--max-num-seqs 64 \
--quantization "wint4" \
--gpu-memory-utilization 0.95 \
--load-choices "default"|2.3.0|
-|ERNIE-4.5-300B-A47B|128K|WINT4|8 (Recommended)|export XPU_VISIBLE_DEVICES="0,1,2,3,4,5,6,7"
python -m fastdeploy.entrypoints.openai.api_server \
--model PaddlePaddle/ERNIE-4.5-300B-A47B-Paddle \
--port 8188 \
--tensor-parallel-size 8 \
--max-model-len 131072 \
--max-num-seqs 64 \
--quantization "wint4" \
--gpu-memory-utilization 0.9 \
--load-choices "default"|2.3.0|
+|ERNIE-4.5-300B-A47B|128K|WINT4|8|export XPU_VISIBLE_DEVICES="0,1,2,3,4,5,6,7"
python -m fastdeploy.entrypoints.openai.api_server \
--model PaddlePaddle/ERNIE-4.5-300B-A47B-Paddle \
--port 8188 \
--tensor-parallel-size 8 \
--max-model-len 131072 \
--max-num-seqs 64 \
--quantization "wint4" \
--gpu-memory-utilization 0.9 \
--load-choices "default"|2.3.0|
|ERNIE-4.5-21B-A3B|32K|BF16|1|export XPU_VISIBLE_DEVICES="0" # Specify any card
python -m fastdeploy.entrypoints.openai.api_server \
--model PaddlePaddle/ERNIE-4.5-21B-A3B-Paddle \
--port 8188 \
--tensor-parallel-size 1 \
--max-model-len 32768 \
--max-num-seqs 128 \
--gpu-memory-utilization 0.9 \
--load-choices "default"|2.3.0|
|ERNIE-4.5-21B-A3B|32K|WINT8|1|export XPU_VISIBLE_DEVICES="0" # Specify any card
python -m fastdeploy.entrypoints.openai.api_server \
--model PaddlePaddle/ERNIE-4.5-21B-A3B-Paddle \
--port 8188 \
--tensor-parallel-size 1 \
--max-model-len 32768 \
--max-num-seqs 128 \
--quantization "wint8" \
--gpu-memory-utilization 0.9 \
--load-choices "default"|2.3.0|
-|ERNIE-4.5-21B-A3B|32K|WINT4|1|export XPU_VISIBLE_DEVICES="0" # Specify any card
python -m fastdeploy.entrypoints.openai.api_server \
--model PaddlePaddle/ERNIE-4.5-21B-A3B-Paddle \
--port 8188 \
--tensor-parallel-size 1 \
--max-model-len 32768 \
--max-num-seqs 128 \
--quantization "wint4" \
--gpu-memory-utilization 0.9 \
--load-choices "default"|2.3.0|
+|ERNIE-4.5-21B-A3B|32K|WINT4|1 (Recommended)|export XPU_VISIBLE_DEVICES="0" # Specify any card
python -m fastdeploy.entrypoints.openai.api_server \
--model PaddlePaddle/ERNIE-4.5-21B-A3B-Paddle \
--port 8188 \
--tensor-parallel-size 1 \
--max-model-len 32768 \
--max-num-seqs 128 \
--quantization "wint4" \
--gpu-memory-utilization 0.9 \
--load-choices "default"|2.3.0|
|ERNIE-4.5-21B-A3B|128K|BF16|1|export XPU_VISIBLE_DEVICES="0" # Specify any card
python -m fastdeploy.entrypoints.openai.api_server \
--model PaddlePaddle/ERNIE-4.5-21B-A3B-Paddle \
--port 8188 \
--tensor-parallel-size 1 \
--max-model-len 131072 \
--max-num-seqs 128 \
--gpu-memory-utilization 0.9 \
--load-choices "default"|2.3.0|
|ERNIE-4.5-21B-A3B|128K|WINT8|1|export XPU_VISIBLE_DEVICES="0" # Specify any card
python -m fastdeploy.entrypoints.openai.api_server \
--model PaddlePaddle/ERNIE-4.5-21B-A3B-Paddle \
--port 8188 \
--tensor-parallel-size 1 \
--max-model-len 131072 \
--max-num-seqs 128 \
--quantization "wint8" \
--gpu-memory-utilization 0.9 \
--load-choices "default"|2.3.0|
-|ERNIE-4.5-21B-A3B|128K|WINT4|1|export XPU_VISIBLE_DEVICES="0" # Specify any card
python -m fastdeploy.entrypoints.openai.api_server \
--model PaddlePaddle/ERNIE-4.5-21B-A3B-Paddle \
--port 8188 \
--tensor-parallel-size 1 \
--max-model-len 131072 \
--max-num-seqs 128 \
--quantization "wint4" \
--gpu-memory-utilization 0.9 \
--load-choices "default"|2.3.0|
+|ERNIE-4.5-21B-A3B|128K|WINT4|1 (Recommended)|export XPU_VISIBLE_DEVICES="0" # Specify any card
python -m fastdeploy.entrypoints.openai.api_server \
--model PaddlePaddle/ERNIE-4.5-21B-A3B-Paddle \
--port 8188 \
--tensor-parallel-size 1 \
--max-model-len 131072 \
--max-num-seqs 128 \
--quantization "wint4" \
--gpu-memory-utilization 0.9 \
--load-choices "default"|2.3.0|
|ERNIE-4.5-0.3B|32K|BF16|1|export XPU_VISIBLE_DEVICES="0" # Specify any card
python -m fastdeploy.entrypoints.openai.api_server \
--model PaddlePaddle/ERNIE-4.5-0.3B-Paddle \
--port 8188 \
--tensor-parallel-size 1 \
--max-model-len 32768 \
--max-num-seqs 128 \
--gpu-memory-utilization 0.9 \
--load-choices "default"|2.3.0|
-|ERNIE-4.5-0.3B|32K|WINT8|1|export XPU_VISIBLE_DEVICES="0" # Specify any card
python -m fastdeploy.entrypoints.openai.api_server \
--model PaddlePaddle/ERNIE-4.5-0.3B-Paddle \
--port 8188 \
--tensor-parallel-size 1 \
--max-model-len 32768 \
--max-num-seqs 128 \
--quantization "wint8" \
--gpu-memory-utilization 0.9 \
--load-choices "default"|2.3.0|
+|ERNIE-4.5-0.3B|32K|WINT8|1 (Recommended)|export XPU_VISIBLE_DEVICES="0" # Specify any card
python -m fastdeploy.entrypoints.openai.api_server \
--model PaddlePaddle/ERNIE-4.5-0.3B-Paddle \
--port 8188 \
--tensor-parallel-size 1 \
--max-model-len 32768 \
--max-num-seqs 128 \
--quantization "wint8" \
--gpu-memory-utilization 0.9 \
--load-choices "default"|2.3.0|
|ERNIE-4.5-0.3B|128K|BF16|1|export XPU_VISIBLE_DEVICES="0" # Specify any card
python -m fastdeploy.entrypoints.openai.api_server \
--model PaddlePaddle/ERNIE-4.5-0.3B-Paddle \
--port 8188 \
--tensor-parallel-size 1 \
--max-model-len 131072 \
--max-num-seqs 128 \
--gpu-memory-utilization 0.9 \
--load-choices "default"|2.3.0|
-|ERNIE-4.5-0.3B|128K|WINT8|1|export XPU_VISIBLE_DEVICES="0" # Specify any card
python -m fastdeploy.entrypoints.openai.api_server \
--model PaddlePaddle/ERNIE-4.5-0.3B-Paddle \
--port 8188 \
--tensor-parallel-size 1 \
--max-model-len 131072 \
--max-num-seqs 128 \
--quantization "wint8" \
--gpu-memory-utilization 0.9 \
--load-choices "default"|2.3.0|
-|ERNIE-4.5-300B-A47B-W4A8C8-TP4|64K|W4A8|4|export XPU_VISIBLE_DEVICES="0,1,2,3" or "4,5,6,7"
python -m fastdeploy.entrypoints.openai.api_server \
--model PaddlePaddle/ERNIE-4.5-300B-A47B-W4A8C8-TP4-Paddle \
--port 8188 \
--tensor-parallel-size 4 \
--max-model-len 32768 \
--max-num-seqs 64 \
--quantization "W4A8" \
--gpu-memory-utilization 0.9 \
--load-choices "default"|2.3.0|
+|ERNIE-4.5-0.3B|128K|WINT8|1 (Recommended)|export XPU_VISIBLE_DEVICES="0" # Specify any card
python -m fastdeploy.entrypoints.openai.api_server \
--model PaddlePaddle/ERNIE-4.5-0.3B-Paddle \
--port 8188 \
--tensor-parallel-size 1 \
--max-model-len 131072 \
--max-num-seqs 128 \
--quantization "wint8" \
--gpu-memory-utilization 0.9 \
--load-choices "default"|2.3.0|
+|ERNIE-4.5-300B-A47B-W4A8C8-TP4|32K|W4A8|4|export XPU_VISIBLE_DEVICES="0,1,2,3" or "4,5,6,7"
python -m fastdeploy.entrypoints.openai.api_server \
--model PaddlePaddle/ERNIE-4.5-300B-A47B-W4A8C8-TP4-Paddle \
--port 8188 \
--tensor-parallel-size 4 \
--max-model-len 32768 \
--max-num-seqs 64 \
--quantization "W4A8" \
--gpu-memory-utilization 0.9 \
--load-choices "default"|2.3.0|
|ERNIE-4.5-VL-28B-A3B|32K|WINT8|1|export XPU_VISIBLE_DEVICES="0" # Specify any card
python -m fastdeploy.entrypoints.openai.api_server \
--model PaddlePaddle/ERNIE-4.5-VL-28B-A3B-Paddle \
--port 8188 \
--tensor-parallel-size 1 \
--quantization "wint8" \
--max-model-len 32768 \
--max-num-seqs 10 \
--enable-mm \
--mm-processor-kwargs '{"video_max_frames": 30}' \
--limit-mm-per-prompt '{"image": 10, "video": 3}' \
--reasoning-parser ernie-45-vl \
--load-choices "default"|2.3.0|
|ERNIE-4.5-VL-424B-A47B|32K|WINT8|8|export XPU_VISIBLE_DEVICES="0,1,2,3,4,5,6,7"
python -m fastdeploy.entrypoints.openai.api_server \
--model PaddlePaddle/ERNIE-4.5-VL-424B-A47B-Paddle \
--port 8188 \
--tensor-parallel-size 8 \
--quantization "wint8" \
--max-model-len 32768 \
--max-num-seqs 8 \
--enable-mm \
--mm-processor-kwargs '{"video_max_frames": 30}' \
--limit-mm-per-prompt '{"image": 10, "video": 3}' \
--reasoning-parser ernie-45-vl \
--load-choices "default"|2.3.0|
|PaddleOCR-VL-0.9B|32K|BF16|1|export FD_ENABLE_MAX_PREFILL=1
export XPU_VISIBLE_DEVICES="0" # Specify any card
python -m fastdeploy.entrypoints.openai.api_server \
--model PaddlePaddle/PaddleOCR-VL \
--port 8188 \
--metrics-port 8181 \
--engine-worker-queue-port 8182 \
--max-model-len 16384 \
--max-num-batched-tokens 16384 \
--gpu-memory-utilization 0.8 \
--max-num-seqs 256|2.3.0|
+|ERNIE-4.5-VL-28B-A3B-Thinking|128K|WINT8|1|export XPU_VISIBLE_DEVICES="0" # Specify any card
python -m fastdeploy.entrypoints.openai.api_server \
--model PaddlePaddle/ERNIE-4.5-VL-28B-A3B-Thinking \
--port 8188 \
--tensor-parallel-size 1 \
--quantization "wint8" \
--max-model-len 131072 \
--max-num-seqs 32 \
--engine-worker-queue-port 8189 \
--metrics-port 8190 \
--cache-queue-port 8191 \
--reasoning-parser ernie-45-vl-thinking \
--tool-call-parser ernie-45-vl-thinking \
--mm-processor-kwargs \ '{"image_max_pixels": 12845056 }' \
--load-choices "default_v1"|2.3.0|
## Quick start
@@ -103,17 +104,17 @@ Deploy the ERNIE-4.5-VL-28B-A3B-Paddle model with WINT8 precision and 32K contex
```bash
export XPU_VISIBLE_DEVICES="0" # Specify any card
python -m fastdeploy.entrypoints.openai.api_server \
---model PaddlePaddle/ERNIE-4.5-VL-28B-A3B-Paddle \
---port 8188 \
---tensor-parallel-size 1 \
---quantization "wint8" \
---max-model-len 32768 \
---max-num-seqs 10 \
---enable-mm \
---mm-processor-kwargs '{"video_max_frames": 30}' \
---limit-mm-per-prompt '{"image": 10, "video": 3}' \
---reasoning-parser ernie-45-vl \
---load-choices "default"
+ --model PaddlePaddle/ERNIE-4.5-VL-28B-A3B-Paddle \
+ --port 8188 \
+ --tensor-parallel-size 1 \
+ --quantization "wint8" \
+ --max-model-len 32768 \
+ --max-num-seqs 10 \
+ --enable-mm \
+ --mm-processor-kwargs '{"video_max_frames": 30}' \
+ --limit-mm-per-prompt '{"image": 10, "video": 3}' \
+ --reasoning-parser ernie-45-vl \
+ --load-choices "default"
```
#### Send requests
@@ -125,7 +126,7 @@ curl -X POST "http://0.0.0.0:8188/v1/chat/completions" \
"messages": [
{"role": "user", "content": [
{"type": "image_url", "image_url": {"url": "https://paddlenlp.bj.bcebos.com/datasets/paddlemix/demo_images/example2.jpg", "detail": "high"}},
- {"type": "text", "text": "请描述图片内容"}
+ {"type": "text", "text": "Please describe the content of the image"}
]}
],
"metadata": {"enable_thinking": false}
@@ -144,7 +145,7 @@ response = client.chat.completions.create(
messages=[
{"role": "user", "content": [
{"type": "image_url", "image_url": {"url": "https://paddlenlp.bj.bcebos.com/datasets/paddlemix/demo_images/example2.jpg", "detail": "high"}},
- {"type": "text", "text": "请描述图片内容"}
+ {"type": "text", "text": "Please describe the content of the image"}
]
},
],
@@ -237,3 +238,77 @@ for chunk in response:
print(reasoning_content + content, end='', flush=True)
print('\n')
```
+
+### Deploy online serving based on ERNIE-4.5-VL-28B-A3B-Thinking
+
+#### Start service
+Deploy the ERNIE-4.5-VL-28B-A3B-Thinking model with WINT8 precision and 128K context length on 1 XPU
+
+```bash
+export XPU_VISIBLE_DEVICES="0"# Specify any card
+python -m fastdeploy.entrypoints.openai.api_server \
+--model PaddlePaddle/ERNIE-4.5-VL-28B-A3B-Thinking \
+--port 8188 \
+--tensor-parallel-size 1 \
+--quantization "wint8" \
+--max-model-len 131072 \
+--max-num-seqs 32 \
+--engine-worker-queue-port 8189 \
+--metrics-port 8190 \
+--cache-queue-port 8191 \
+--reasoning-parser ernie-45-vl-thinking \
+--tool-call-parser ernie-45-vl-thinking \
+--mm-processor-kwargs '{"image_max_pixels": 12845056 }' \
+--load-choices "default_v1"
+```
+
+#### Send requests
+
+```bash
+curl -X POST "http://0.0.0.0:8188/v1/chat/completions" \
+-H "Content-Type: application/json" \
+-d '{
+ "messages": [
+ {"role": "user", "content": [
+ {"type": "image_url", "image_url": {"url": "https://paddlenlp.bj.bcebos.com/datasets/paddlemix/demo_images/example2.jpg", "detail": "high"}},
+ {"type": "text", "text": "Please describe the content of the image"}
+ ]}
+ ],
+ "metadata": {"enable_thinking": true}
+}'
+```
+
+```python
+import openai
+
+ip = "0.0.0.0"
+service_http_port = "8188"
+client = openai.Client(base_url=f"http://{ip}:{service_http_port}/v1", api_key="EMPTY_API_KEY")
+
+response = client.chat.completions.create(
+ model="default",
+ messages=[
+ {"role": "user", "content": [
+ {"type": "image_url", "image_url": {"url": "https://paddlenlp.bj.bcebos.com/datasets/paddlemix/demo_images/example2.jpg", "detail": "high"}},
+ {"type": "text", "text": "Please describe the content of the image"}
+ ]
+ },
+ ],
+ temperature=0.0001,
+ max_tokens=10000,
+ stream=True,
+ top_p=0,
+ metadata={"enable_thinking": True},
+)
+
+def get_str(content_raw):
+ content_str = str(content_raw) if content_raw is not None else ''
+ return content_str
+
+for chunk in response:
+ if chunk.choices[0].delta is not None and chunk.choices[0].delta.role != 'assistant':
+ reasoning_content = get_str(chunk.choices[0].delta.reasoning_content)
+ content = get_str(chunk.choices[0].delta.content)
+ print(reasoning_content + content + is_reason, end='', flush=True)
+print('\n')
+```
diff --git a/docs/zh/usage/kunlunxin_xpu_deployment.md b/docs/zh/usage/kunlunxin_xpu_deployment.md
index b85b48f92..b6042e13e 100644
--- a/docs/zh/usage/kunlunxin_xpu_deployment.md
+++ b/docs/zh/usage/kunlunxin_xpu_deployment.md
@@ -6,21 +6,22 @@
|ERNIE-4.5-300B-A47B|32K|WINT8|8|export XPU_VISIBLE_DEVICES="0,1,2,3,4,5,6,7"
python -m fastdeploy.entrypoints.openai.api_server \
--model PaddlePaddle/ERNIE-4.5-300B-A47B-Paddle \
--port 8188 \
--tensor-parallel-size 8 \
--max-model-len 32768 \
--max-num-seqs 64 \
--quantization "wint8" \
--gpu-memory-utilization 0.9 \
--load-choices "default"|2.3.0|
|ERNIE-4.5-300B-A47B|32K|WINT4|4 （推荐）|export XPU_VISIBLE_DEVICES="0,1,2,3" or "4,5,6,7"
python -m fastdeploy.entrypoints.openai.api_server \
--model PaddlePaddle/ERNIE-4.5-300B-A47B-Paddle \
--port 8188 \
--tensor-parallel-size 4 \
--max-model-len 32768 \
--max-num-seqs 64 \
--quantization "wint4" \
--gpu-memory-utilization 0.9 \
--load-choices "default"|2.3.0|
|ERNIE-4.5-300B-A47B|32K|WINT4|8|export XPU_VISIBLE_DEVICES="0,1,2,3,4,5,6,7"
python -m fastdeploy.entrypoints.openai.api_server \
--model PaddlePaddle/ERNIE-4.5-300B-A47B-Paddle \
--port 8188 \
--tensor-parallel-size 8 \
--max-model-len 32768 \
--max-num-seqs 64 \
--quantization "wint4" \
--gpu-memory-utilization 0.95 \
--load-choices "default"|2.3.0|
-|ERNIE-4.5-300B-A47B|128K|WINT4|8 （推荐）|export XPU_VISIBLE_DEVICES="0,1,2,3,4,5,6,7"
python -m fastdeploy.entrypoints.openai.api_server \
--model PaddlePaddle/ERNIE-4.5-300B-A47B-Paddle \
--port 8188 \
--tensor-parallel-size 8 \
--max-model-len 131072 \
--max-num-seqs 64 \
--quantization "wint4" \
--gpu-memory-utilization 0.9 \
--load-choices "default"|2.3.0|
+|ERNIE-4.5-300B-A47B|128K|WINT4|8|export XPU_VISIBLE_DEVICES="0,1,2,3,4,5,6,7"
python -m fastdeploy.entrypoints.openai.api_server \
--model PaddlePaddle/ERNIE-4.5-300B-A47B-Paddle \
--port 8188 \
--tensor-parallel-size 8 \
--max-model-len 131072 \
--max-num-seqs 64 \
--quantization "wint4" \
--gpu-memory-utilization 0.9 \
--load-choices "default"|2.3.0|
|ERNIE-4.5-21B-A3B|32K|BF16|1|export XPU_VISIBLE_DEVICES="0" # 指定任意一张卡
python -m fastdeploy.entrypoints.openai.api_server \
--model PaddlePaddle/ERNIE-4.5-21B-A3B-Paddle \
--port 8188 \
--tensor-parallel-size 1 \
--max-model-len 32768 \
--max-num-seqs 128 \
--gpu-memory-utilization 0.9 \
--load-choices "default"|2.3.0|
|ERNIE-4.5-21B-A3B|32K|WINT8|1|export XPU_VISIBLE_DEVICES="0" # 指定任意一张卡
python -m fastdeploy.entrypoints.openai.api_server \
--model PaddlePaddle/ERNIE-4.5-21B-A3B-Paddle \
--port 8188 \
--tensor-parallel-size 1 \
--max-model-len 32768 \
--max-num-seqs 128 \
--quantization "wint8" \
--gpu-memory-utilization 0.9 \
--load-choices "default"|2.3.0|
-|ERNIE-4.5-21B-A3B|32K|WINT4|1|export XPU_VISIBLE_DEVICES="0" # 指定任意一张卡
python -m fastdeploy.entrypoints.openai.api_server \
--model PaddlePaddle/ERNIE-4.5-21B-A3B-Paddle \
--port 8188 \
--tensor-parallel-size 1 \
--max-model-len 32768 \
--max-num-seqs 128 \
--quantization "wint4" \
--gpu-memory-utilization 0.9 \
--load-choices "default"|2.3.0|
+|ERNIE-4.5-21B-A3B|32K|WINT4|1 （推荐）|export XPU_VISIBLE_DEVICES="0" # 指定任意一张卡
python -m fastdeploy.entrypoints.openai.api_server \
--model PaddlePaddle/ERNIE-4.5-21B-A3B-Paddle \
--port 8188 \
--tensor-parallel-size 1 \
--max-model-len 32768 \
--max-num-seqs 128 \
--quantization "wint4" \
--gpu-memory-utilization 0.9 \
--load-choices "default"|2.3.0|
|ERNIE-4.5-21B-A3B|128K|BF16|1|export XPU_VISIBLE_DEVICES="0" # 指定任意一张卡
python -m fastdeploy.entrypoints.openai.api_server \
--model PaddlePaddle/ERNIE-4.5-21B-A3B-Paddle \
--port 8188 \
--tensor-parallel-size 1 \
--max-model-len 131072 \
--max-num-seqs 128 \
--gpu-memory-utilization 0.9 \
--load-choices "default"|2.3.0|
|ERNIE-4.5-21B-A3B|128K|WINT8|1|export XPU_VISIBLE_DEVICES="0" # 指定任意一张卡
python -m fastdeploy.entrypoints.openai.api_server \
--model PaddlePaddle/ERNIE-4.5-21B-A3B-Paddle \
--port 8188 \
--tensor-parallel-size 1 \
--max-model-len 131072 \
--max-num-seqs 128 \
--quantization "wint8" \
--gpu-memory-utilization 0.9 \
--load-choices "default"|2.3.0|
-|ERNIE-4.5-21B-A3B|128K|WINT4|1|export XPU_VISIBLE_DEVICES="0" # 指定任意一张卡
python -m fastdeploy.entrypoints.openai.api_server \
--model PaddlePaddle/ERNIE-4.5-21B-A3B-Paddle \
--port 8188 \
--tensor-parallel-size 1 \
--max-model-len 131072 \
--max-num-seqs 128 \
--quantization "wint4" \
--gpu-memory-utilization 0.9 \
--load-choices "default"|2.3.0|
+|ERNIE-4.5-21B-A3B|128K|WINT4|1 （推荐）|export XPU_VISIBLE_DEVICES="0" # 指定任意一张卡
python -m fastdeploy.entrypoints.openai.api_server \
--model PaddlePaddle/ERNIE-4.5-21B-A3B-Paddle \
--port 8188 \
--tensor-parallel-size 1 \
--max-model-len 131072 \
--max-num-seqs 128 \
--quantization "wint4" \
--gpu-memory-utilization 0.9 \
--load-choices "default"|2.3.0|
|ERNIE-4.5-0.3B|32K|BF16|1|export XPU_VISIBLE_DEVICES="0" # 指定任意一张卡
python -m fastdeploy.entrypoints.openai.api_server \
--model PaddlePaddle/ERNIE-4.5-0.3B-Paddle \
--port 8188 \
--tensor-parallel-size 1 \
--max-model-len 32768 \
--max-num-seqs 128 \
--gpu-memory-utilization 0.9 \
--load-choices "default"|2.3.0|
-|ERNIE-4.5-0.3B|32K|WINT8|1|export XPU_VISIBLE_DEVICES="x" # 指定任意一张卡
python -m fastdeploy.entrypoints.openai.api_server \
--model PaddlePaddle/ERNIE-4.5-0.3B-Paddle \
--port 8188 \
--tensor-parallel-size 1 \
--max-model-len 32768 \
--max-num-seqs 128 \
--quantization "wint8" \
--gpu-memory-utilization 0.9 \
--load-choices "default"|2.3.0|
+|ERNIE-4.5-0.3B|32K|WINT8|1 （推荐）|export XPU_VISIBLE_DEVICES="x" # 指定任意一张卡
python -m fastdeploy.entrypoints.openai.api_server \
--model PaddlePaddle/ERNIE-4.5-0.3B-Paddle \
--port 8188 \
--tensor-parallel-size 1 \
--max-model-len 32768 \
--max-num-seqs 128 \
--quantization "wint8" \
--gpu-memory-utilization 0.9 \
--load-choices "default"|2.3.0|
|ERNIE-4.5-0.3B|128K|BF16|1|export XPU_VISIBLE_DEVICES="0" # 指定任意一张卡
python -m fastdeploy.entrypoints.openai.api_server \
--model PaddlePaddle/ERNIE-4.5-0.3B-Paddle \
--port 8188 \
--tensor-parallel-size 1 \
--max-model-len 131072 \
--max-num-seqs 128 \
--gpu-memory-utilization 0.9 \
--load-choices "default"|2.3.0|
-|ERNIE-4.5-0.3B|128K|WINT8|1|export XPU_VISIBLE_DEVICES="0" # 指定任意一张卡
python -m fastdeploy.entrypoints.openai.api_server \
--model PaddlePaddle/ERNIE-4.5-0.3B-Paddle \
--port 8188 \
--tensor-parallel-size 1 \
--max-model-len 131072 \
--max-num-seqs 128 \
--quantization "wint8" \
--gpu-memory-utilization 0.9 \
--load-choices "default"|2.3.0|
-|ERNIE-4.5-300B-A47B-W4A8C8-TP4|64K|W4A8|4|export XPU_VISIBLE_DEVICES="0,1,2,3" or "4,5,6,7"
python -m fastdeploy.entrypoints.openai.api_server \
--model PaddlePaddle/ERNIE-4.5-300B-A47B-W4A8C8-TP4-Paddle \
--port 8188 \
--tensor-parallel-size 4 \
--max-model-len 32768 \
--max-num-seqs 64 \
--quantization "W4A8" \
--gpu-memory-utilization 0.9 \
--load-choices "default"|2.3.0|
+|ERNIE-4.5-0.3B|128K|WINT8|1 （推荐）|export XPU_VISIBLE_DEVICES="0" # 指定任意一张卡
python -m fastdeploy.entrypoints.openai.api_server \
--model PaddlePaddle/ERNIE-4.5-0.3B-Paddle \
--port 8188 \
--tensor-parallel-size 1 \
--max-model-len 131072 \
--max-num-seqs 128 \
--quantization "wint8" \
--gpu-memory-utilization 0.9 \
--load-choices "default"|2.3.0|
+|ERNIE-4.5-300B-A47B-W4A8C8-TP4|32K|W4A8|4|export XPU_VISIBLE_DEVICES="0,1,2,3" or "4,5,6,7"
python -m fastdeploy.entrypoints.openai.api_server \
--model PaddlePaddle/ERNIE-4.5-300B-A47B-W4A8C8-TP4-Paddle \
--port 8188 \
--tensor-parallel-size 4 \
--max-model-len 32768 \
--max-num-seqs 64 \
--quantization "W4A8" \
--gpu-memory-utilization 0.9 \
--load-choices "default"|2.3.0|
|ERNIE-4.5-VL-28B-A3B|32K|WINT8|1|export XPU_VISIBLE_DEVICES="0"# 指定任意一张卡
python -m fastdeploy.entrypoints.openai.api_server \
--model PaddlePaddle/ERNIE-4.5-VL-28B-A3B-Paddle \
--port 8188 \
--tensor-parallel-size 1 \
--quantization "wint8" \
--max-model-len 32768 \
--max-num-seqs 10 \
--enable-mm \
--mm-processor-kwargs '{"video_max_frames": 30}' \
--limit-mm-per-prompt '{"image": 10, "video": 3}' \
--reasoning-parser ernie-45-vl \
--load-choices "default"|2.3.0|
|ERNIE-4.5-VL-424B-A47B|32K|WINT8|8|export XPU_VISIBLE_DEVICES="0,1,2,3,4,5,6,7"
python -m fastdeploy.entrypoints.openai.api_server \
--model PaddlePaddle/ERNIE-4.5-VL-424B-A47B-Paddle \
--port 8188 \
--tensor-parallel-size 8 \
--quantization "wint8" \
--max-model-len 32768 \
--max-num-seqs 8 \
--enable-mm \
--mm-processor-kwargs '{"video_max_frames": 30}' \
--limit-mm-per-prompt '{"image": 10, "video": 3}' \
--reasoning-parser ernie-45-vl \
--load-choices "default"|2.3.0|
|PaddleOCR-VL-0.9B|32K|BF16|1|export FD_ENABLE_MAX_PREFILL=1
export XPU_VISIBLE_DEVICES="0" # 指定任意一张卡
python -m fastdeploy.entrypoints.openai.api_server \
--model PaddlePaddle/PaddleOCR-VL \
--port 8188 \
--metrics-port 8181 \
--engine-worker-queue-port 8182 \
--max-model-len 16384 \
--max-num-batched-tokens 16384 \
--gpu-memory-utilization 0.8 \
--max-num-seqs 256|2.3.0|
+|ERNIE-4.5-VL-28B-A3B-Thinking|128K|WINT8|1|export XPU_VISIBLE_DEVICES="0" # 指定任意一张卡
python -m fastdeploy.entrypoints.openai.api_server \
--model PaddlePaddle/ERNIE-4.5-VL-28B-A3B-Thinking \
--port 8188 \
--tensor-parallel-size 1 \
--quantization "wint8" \
--max-model-len 131072 \
--max-num-seqs 32 \
--engine-worker-queue-port 8189 \
--metrics-port 8190 \
--cache-queue-port 8191 \
--reasoning-parser ernie-45-vl-thinking \
--tool-call-parser ernie-45-vl-thinking \
--mm-processor-kwargs \ '{"image_max_pixels": 12845056 }' \
--load-choices "default_v1"|2.3.0|
## 快速开始
@@ -103,19 +104,19 @@ OpenAI 协议的更多说明可参考文档 [OpenAI Chat Completion API](https:/
基于 WINT8 精度和 32K 上下文部署 ERNIE-4.5-VL-28B-A3B-Paddle 模型到 单卡 P800 服务器
```bash
-export XPU_VISIBLE_DEVICES="0" # Specify any card
+export XPU_VISIBLE_DEVICES="0" # 指定任意一张卡
python -m fastdeploy.entrypoints.openai.api_server \
---model PaddlePaddle/ERNIE-4.5-VL-28B-A3B-Paddle \
---port 8188 \
---tensor-parallel-size 1 \
---quantization "wint8" \
---max-model-len 32768 \
---max-num-seqs 10 \
---enable-mm \
---mm-processor-kwargs '{"video_max_frames": 30}' \
---limit-mm-per-prompt '{"image": 10, "video": 3}' \
---reasoning-parser ernie-45-vl \
---load-choices "default"
+ --model PaddlePaddle/ERNIE-4.5-VL-28B-A3B-Paddle \
+ --port 8188 \
+ --tensor-parallel-size 1 \
+ --quantization "wint8" \
+ --max-model-len 32768 \
+ --max-num-seqs 10 \
+ --enable-mm \
+ --mm-processor-kwargs '{"video_max_frames": 30}' \
+ --limit-mm-per-prompt '{"image": 10, "video": 3}' \
+ --reasoning-parser ernie-45-vl \
+ --load-choices "default"
```
#### 请求服务
@@ -239,3 +240,78 @@ for chunk in response:
print(reasoning_content + content, end='', flush=True)
print('\n')
```
+
+### 基于ERNIE-4.5-VL-28B-A3B-Thinking模型部署在线服务
+
+#### 启动服务
+
+基于 WINT8 精度和 128K 上下文部署 ERNIE-4.5-VL-28B-A3B-Thinking 模型到 单卡 P800 服务器
+
+```bash
+export XPU_VISIBLE_DEVICES="0"# 指定任意一张卡
+python -m fastdeploy.entrypoints.openai.api_server \
+ --model PaddlePaddle/ERNIE-4.5-VL-28B-A3B-Thinking \
+ --port 8188 \
+ --tensor-parallel-size 1 \
+ --quantization "wint8" \
+ --max-model-len 131072 \
+ --max-num-seqs 32 \
+ --engine-worker-queue-port 8189 \
+ --metrics-port 8190 \
+ --cache-queue-port 8191 \
+ --reasoning-parser ernie-45-vl-thinking \
+ --tool-call-parser ernie-45-vl-thinking \
+ --mm-processor-kwargs '{"image_max_pixels": 12845056 }' \
+ --load-choices "default_v1"
+```
+
+#### 请求服务
+
+```bash
+curl -X POST "http://0.0.0.0:8188/v1/chat/completions" \
+-H "Content-Type: application/json" \
+-d '{
+ "messages": [
+ {"role": "user", "content": [
+ {"type": "image_url", "image_url": {"url": "https://paddlenlp.bj.bcebos.com/datasets/paddlemix/demo_images/example2.jpg", "detail": "high"}},
+ {"type": "text", "text": "请描述图片内容"}
+ ]}
+ ],
+ "metadata": {"enable_thinking": true}
+}'
+```
+
+```python
+import openai
+
+ip = "0.0.0.0"
+service_http_port = "8188"
+client = openai.Client(base_url=f"http://{ip}:{service_http_port}/v1", api_key="EMPTY_API_KEY")
+
+response = client.chat.completions.create(
+ model="default",
+ messages=[
+ {"role": "user", "content": [
+ {"type": "image_url", "image_url": {"url": "https://paddlenlp.bj.bcebos.com/datasets/paddlemix/demo_images/example2.jpg", "detail": "high"}},
+ {"type": "text", "text": "请描述图片内容"}
+ ]
+ },
+ ],
+ temperature=0.0001,
+ max_tokens=10000,
+ stream=True,
+ top_p=0,
+ metadata={"enable_thinking": True},
+)
+
+def get_str(content_raw):
+ content_str = str(content_raw) if content_raw is not None else ''
+ return content_str
+
+for chunk in response:
+ if chunk.choices[0].delta is not None and chunk.choices[0].delta.role != 'assistant':
+ reasoning_content = get_str(chunk.choices[0].delta.reasoning_content)
+ content = get_str(chunk.choices[0].delta.content)
+ print(reasoning_content + content, end='', flush=True)
+print('\n')
+```