EthanYe

Fix quantize model deploy bugs when using MKLDNN

1

### PR types Bug fixes ### PR changes Others ### Describe Fix ppyoloe、MobileNetV3_large、Ernie 3.0-Medium and bert acc drop bugs. 1. Avoid the problem that the weights are quantize and dequantized...

1. The quantization cache file read by TensorRT is returned by string 2. remove scale file path set

### PR types(PR类型) Backend ### Describe 通过PaddleLite的芯原 TIM-VX支持Rockchip RV1126 Soc的部署。 mobilenetv1全量化模型 NPU avg latency：6.12 ms CPU avg latency：331.93 ms

### Describe - Update p2o version to 1.0.3

### PR types(PR类型) Backend ### Describe Add YOLOv5、PPYOLOE and PP-Liteseg for RV1126

[Backend] Add KunlunXin XPU deploy support

1

### PR types(PR类型) Backend ### Describe 1. Add KunlunXin XPU deploy support 2. Add paddleclas support 3. Add Yolov5 support

### PR types(PR类型) Backend ### Description Update P2O to 1.0.5

[Backend] A311D support

1

### PR types(PR类型) Backend ### Description A311D support ResNet50 MobileNet PP-YOLOE YOLOv5 PP-LiteSeg

1. Convert paddle model to standard model ``` python convert_paddle_to_standard.py --paddle_model MobileNetV3/inference --save_dir test_model/ ``` 2. Process standard model, include convert to paddle model, get model、graph、node、variable_type、tensors ``` python process_standard_model.py --standard_model...

[BUG] chatglm-6b模型推理服务报错

5

lightllm commit id：718e6d6dfffc75e7bbfd7ea80ba4afb77aa27726 huggingfaced下载的chatglm-6b模型，启动服务的时候报错。 模型下载链接：https://huggingface.co/THUDM/chatglm-6b 服务启动命令：python -m lightllm.server.api_server --model_dir THUDM/chatglm-6b --host 0.0.0.0 --port 8100 --tp 1 --max_total_token_num 120000 --tokenizer_mode auto --trust_remote_code 报错信息： ################ load model error: 'ffn_hidden_size' 'ffn_hidden_size' File "/lightllm/lightllm/models/chatglm2/layer_weights/transformer_layer_weight.py",...