From 4e0aae556bbfb81f765ddb3e247f34441c607c5e Mon Sep 17 00:00:00 2001
From: 游雁 <zhifu.gzf@alibaba-inc.com>
Date: 星期五, 21 四月 2023 10:45:16 +0800
Subject: [PATCH] docs
---
funasr/export/README.md | 34 ++++++++++++++++++++++++++--------
1 files changed, 26 insertions(+), 8 deletions(-)
diff --git a/funasr/export/README.md b/funasr/export/README.md
index bde1e94..d403121 100644
--- a/funasr/export/README.md
+++ b/funasr/export/README.md
@@ -1,13 +1,20 @@
+# Export models
## Environments
- torch >= 1.11.0
- modelscope >= 1.2.0
- torch-quant >= 0.4.0 (required for exporting quantized torchscript format model)
- # pip install torch-quant -i https://pypi.org/simple
+### Install modelscope and funasr
-## Install modelscope and funasr
-
-The installation is the same as [funasr](../../README.md)
+The installation is the same as [funasr](https://github.com/alibaba-damo-academy/FunASR/blob/main/README.md#installation)
+```shell
+# pip3 install torch torchaudio
+pip install -U modelscope funasr
+# For the users in China, you could install with the command:
+# pip install -U modelscope funasr -i https://mirror.sjtu.edu.cn/pypi/web/simple
+```
+### Install the quantization tools
+```shell
+pip install torch-quant # Optional, for torchscript quantization
+pip install onnx onnxruntime # Optional, for onnx quantization
+```
## Export model
`Tips`: torch>=1.11.0
@@ -17,7 +24,7 @@
--model-name [model_name] \
--export-dir [export_dir] \
--type [onnx, torch] \
- --quantize \
+ --quantize [true, false] \
--fallback-num [fallback_num]
```
`model-name`: the model is to export. It could be the models from modelscope, or local finetuned model(named: model.pb).
@@ -30,6 +37,16 @@
`fallback-num`: specify the number of fallback layers to perform automatic mixed precision quantization.
+## Performance Benchmark of Runtime
+
+### Paraformer on CPU
+
+[onnx runtime](https://github.com/alibaba-damo-academy/FunASR/blob/main/funasr/runtime/python/benchmark_onnx.md)
+
+[libtorch runtime](https://github.com/alibaba-damo-academy/FunASR/blob/main/funasr/runtime/python/benchmark_libtorch.md)
+
+### Paraformer on GPU
+[nv-triton](https://github.com/alibaba-damo-academy/FunASR/tree/main/funasr/runtime/triton_gpu)
## For example
### Export onnx format model
@@ -55,3 +72,4 @@
## Acknowledge
Torch model quantization is supported by [BladeDISC](https://github.com/alibaba/BladeDISC), an end-to-end DynamIc Shape Compiler project for machine learning workloads. BladeDISC provides general, transparent, and ease of use performance optimization for TensorFlow/PyTorch workloads on GPGPU and CPU backends. If you are interested, please contact us.
+
--
Gitblit v1.9.1