[Request] 希望提供一个类似NextChat的清除聊天功能

[zhufeizzz]

🥰 需求描述

使用场景：
1、大部分会话不需要前置历史对话，附带历史消息会影响模型生成效果，并浪费大量Token。
2、有的时候又需要历史对话，现在Lobe的设计不能很好的支持两种模式的切换

问题1：如果要清除上下文，必须新建会话或清空整个会话，这样会导致出现大量的会话或者历史消息丢失。
问题2：如果想使用历史消息数临时调整范围，每次必须手工选择数字，操作很不方便。

🧐 解决方案

希望提供一个类似NextChat的清除聊天功能
可以在会话内主动清除上下文，但是又不清除整个会话，这样可以方便选择要不要历史会话

📝 补充信息

No response

[lobehubbot]

👀 @zhufeizzz

Thank you for raising an issue. We will investigate into the matter and get back to you as soon as possible.
Please make sure you have given us as much context as possible.
非常感谢您提交 issue。我们会尽快调查此事，并尽快回复您。 请确保您已经提供了尽可能多的背景信息。

[lobehubbot]

Bot detected the issue body's language is not English, translate it automatically. 👯👭🏻🧑‍🤝‍🧑👫🧑🏿‍🤝‍🧑🏻👩🏾‍🤝‍👨🏿👬🏿

---

🥰 Description of requirements

Usage scenarios:

1. Most conversations do not require pre-existing historical conversations. The accompanying historical messages will affect the model generation effect and waste a lot of tokens.
2. Sometimes historical dialogue is needed. The current design of Lobe does not support switching between the two modes very well.

Problem 1: If you want to clear the context, you must create a new session or clear the entire session, which will cause a large number of sessions or historical messages to be lost.
Question 2: If you want to use historical message numbers to temporarily adjust the range, you must manually select the number each time, which is very inconvenient.

🧐 Solution

Hope to provide a clear chat function similar to NextChat
You can proactively clear the context within a session without clearing the entire session. This makes it easy to choose whether to have historical sessions.

📝 Supplementary information

No response

[arvinxx]

1. 现在我们的设计就是鼓励你多开会话。这有什么问题不？
2. Next Chat的清除会话为什么能解决你说的问题？他还能撤销吗？

[lobehubbot]

Bot detected the issue body's language is not English, translate it automatically. 👯👭🏻🧑‍🤝‍🧑👫🧑🏿‍🤝‍🧑🏻👩🏾‍🤝‍👨🏿👬🏿

---

1. Our design now encourages you to have more conversations. Is there any problem with this?

[zhufeizzz]

@arvinxx
1、会话太多了列表里面几百个会话不好找。如果是放在一个会话往上翻就行了，我一般是几天新建一个会话，历史消息很好找。
2、Next Chat是默认会使用历史会话上下文，如果不想保留点一下按钮就行，仅需要一步操作。撤销也支持，再点击一次就行了。

Lobe里面想达到一样的操作要点击好几回。

[lobehubbot]

Bot detected the issue body's language is not English, translate it automatically. 👯👭🏻🧑‍🤝‍🧑👫🧑🏿‍🤝‍🧑🏻👩🏾‍🤝‍👨🏿👬🏿

---

@arvinxx

1. There are too many sessions and it is difficult to find hundreds of sessions in the list. If you put it in a conversation, just scroll up. I usually create a new conversation in a few days, and historical information is easy to find.
2. Next Chat will use the historical conversation context by default. If you don’t want to keep it, just click the button. It only requires one step. Undo is also supported, just click it again. In Lobe, you have to click several times to achieve the same operation.

[arvinxx]

1、会话太多了列表里面几百个会话不好找

针对这个问题，后面会新增话题分组的功能（现在其实是自动根据时间做了分组，后面会支持自定义分组），这样一来就可以很好地去组织相关类型的话题。

如果是放在一个会话往上翻就行了，我一般是几天新建一个会话，历史消息很好找。Next Chat是默认会使用历史会话上下文，如果不想保留点一下按钮就行，仅需要一步操作。撤销也支持，再点击一次就行了。

我觉得你这个操作并没有根本解决你说的浪费 token 的问题。比如你不想保留的时候点了按钮，新聊了10句。但是这个时候如果你要用 10句之前的上下文，你仍然需要撤销清除，同时为了10句前的上下文又得加进去这10句不相干的上下文。token 也并没有省下。

而拆话题的话则能很方便地回到之前聊的那个话题，继续对话。

[lobehubbot]

Bot detected the issue body's language is not English, translate it automatically. 👯👭🏻🧑‍🤝‍🧑👫🧑🏿‍🤝‍🧑🏻👩🏾‍🤝‍👨🏿👬🏿

---

1. There are too many sessions and it is difficult to find hundreds of sessions in the list.

In response to this problem, the function of topic grouping will be added later (now it is automatically grouped according to time, and custom grouping will be supported later), so that related types of topics can be well organized.

If you put it in a conversation, just scroll up. I usually create a new conversation in a few days. Historical messages are easy to find. Next Chat will use the historical conversation context by default. If you don't want to keep it, just click the button. It only requires one step. Undo is also supported, just click it again.

I think your operation does not fundamentally solve the problem of wasting tokens that you mentioned. For example, you clicked the button when you didn’t want to keep it and chatted 10 new sentences. But at this time, if you want to use the context 10 sentences ago, you still need to undo the clearing, and at the same time, you have to add these 10 irrelevant contexts for the context 10 sentences ago. The token is not saved either.

And if you break the topic, you can easily return to the previous topic and continue the conversation.

[arvinxx]

另外查找的诉求根本上可以通过 CMD K 这种快捷操作的统一查询来实现，这个是在明年的 roadmap 里的

[lobehubbot]

Bot detected the issue body's language is not English, translate it automatically. 👯👭🏻🧑‍🤝‍🧑👫🧑🏿‍🤝‍🧑🏻👩🏾‍🤝‍👨🏿👬🏿

---

In addition, the search request can basically be realized through the unified query of CMD K, a quick operation, which will be included in next year’s roadmap.

[zhufeizzz]

我觉得你这个操作并没有根本解决你说的浪费 token 的问题。比如你不想保留的时候点了按钮，新聊了10句。但是这个时候如果你要用 10句之前的上下文，你仍然需要撤销清除，同时为了10句前的上下文又得加进去这10句不相干的上下文。token 也并没有省下。

因为没有你说的这种场景，平时主要问开发设计类问题，当时就会问清楚，想不到有啥场景还要继续之前的对话。

而且即使真有人有这种诉求，新开会话就可以了。这个按钮只是给用户多了一种选择，简化操作。

CherryStudio也有这个功能，应该算是一个比较通用的能力。

[lobehubbot]

Bot detected the issue body's language is not English, translate it automatically. 👯👭🏻🧑‍🤝‍🧑👫🧑🏿‍🤝‍🧑🏻👩🏾‍🤝‍👨🏿👬🏿

---

I think your operation does not fundamentally solve the problem of wasting tokens that you mentioned. For example, you clicked the button when you didn’t want to keep it and chatted 10 new sentences. But at this time, if you want to use the context 10 sentences ago, you still need to undo the clearing, and at the same time, you have to add these 10 irrelevant contexts to the context 10 sentences ago. The token is not saved either.

Because there is no such scenario as you mentioned, I usually ask questions about development and design. I will ask clearly at that time, and I will continue the previous conversation if there is any unexpected scenario.

And even if someone really has this request, just open a new session. This button just gives users one more choice and simplifies operations.