Luo Peng

Problem with path matching and URI manipulation when involving double slashes

4

## Problem ### 1. Path matcher containing double slashes won't match any request Caddyfile: ``` :8080 { handle_path //prefix/* { reverse_proxy 127.0.0.1:9090 } respond "oops" } ``` Neither of the...

## Motivation Implement something like [Timer.Reset](https://golang.org/pkg/time/#Timer.Reset), to reuse an existing timer and reset its duration. ## Solutions There are three candidate solutions: ### 1. Add Timer.Reset ```go func (t *Timer)...

While maintaining lots of actors in a cluster, chances are that we need to gracefully restart a node for version update (similar to [this orleans issue](https://github.com/dotnet/orleans/issues/381)). To gracefully restart a...

Sorry for disturbing you folks, but why is this project deprecated? After googling, I found nothing other than [this twitter comment](https://twitter.com/conormyhrvold/status/1032464629664952320): > Uber has actually moved to using Kafka. Is...

Problem with go-chi/chi

1

### The following program `sample.go` triggers an unexpected result ```go package main import "github.com/go-chi/chi" func main() { router := chi.NewRouter() println("router:", router) } ``` ### Expected result ```console $ go...

Fixed #10.

There are two types of errors: 1. Input errors, which should be treated as **soft failures** - Not found: the input username does not exist - Invalid Credentials: the input...

如何在 Paddle Inference 中使用 RocketQA 的模型？

2

参考 [Paddle Inference 文档](https://www.paddlepaddle.org.cn/inference/master/guides/introduction/workflow.html)： > Paddle Inference 原生支持由 [PaddlePaddle](https://github.com/PaddlePaddle/Paddle) 深度学习框架训练产出的推理模型。PaddlePaddle 用于推理的模型分别可通过 `paddle.jit.save` (动态图) 与 `paddle.static.save_inference_model` (静态图) 或 `paddle.Model().save` (高层API) 保存下来。 目前 RocketQA 使用的是老版本的 PaddlePaddle，用于推理的模型应该需要通过 `fluid.io.save_inference_model` 保存，但目前看代码是通过 `fluid.io.save_persistables` 保存的： https://github.com/PaddlePaddle/RocketQA/blob/019ad5c1088167e264c5ec799c5f7fd22e39ad27/rocketqa/encoder/dual_encoder.py#L380 https://github.com/PaddlePaddle/RocketQA/blob/019ad5c1088167e264c5ec799c5f7fd22e39ad27/rocketqa/encoder/cross_encoder.py#L326...

提供命令行工具用于简化模型训练

3

## 问题 当前训练或微调一个模型，除了构造训练集，还需要做以下多个步骤： 1. 参考 [Train Your Own Model](https://github.com/PaddlePaddle/RocketQA#train-your-own-model) 编写一个训练脚本 2. 训练结束后，需要 [删除其中 *moment* 文件](https://github.com/PaddlePaddle/RocketQA/issues/61)，以减小模型 3. 为了运行模型，需要通过拷贝 [examples/de_models](https://github.com/PaddlePaddle/RocketQA/tree/main/examples/de_models) 或 [examples/ce_models](https://github.com/PaddlePaddle/RocketQA/tree/main/examples/ce_models) 建立模型目录 4. 然后修改 config.json 中的 `${YOUR_MODEL}`，指向第 2 步中的目录（[参考1](https://github.com/PaddlePaddle/RocketQA/issues/33)、[参考2](https://github.com/PaddlePaddle/RocketQA/issues/57#issuecomment-1212667559)） 个人以为，上述训练操作除了步骤较多比较繁琐，还会对初学者造成比较大的困惑（初学者更喜欢开箱即用和一键操作）。 ##...

Add support for the business level input validation

1

## Problem `kun` has already supported request validation since #10. However, there are some disadvantages: - The request validation is done at the transport layer, while the request fields to...