You signed in with another tab or window. Reload to refresh your session.You signed out in another tab or window. Reload to refresh your session.You switched accounts on another tab or window. Reload to refresh your session.Dismiss alert
I don't know how to explain this, and as there are no version numbers to chatgpt-web, I cannot tell you what version I had before the update. I am not familiar with the terminology to properly express myself. I am going to do my best:
The Send Feature is buggy. I thought the update would improve the send feature, but it has only gotten worse for my usecase. I don't know if this is by design and is in fact better for everyone else, or just me who finds it a challenge.
Here are the basic features:
CONDITIONS
Chat Bubbles can be "Greyed out" (ignored)
Chat Bubbles that finish text in the middle of GPT's thought can be continued with the same Chat Bubble
Chat Bubbles that had thought streams finished/completed create a new Chat Bubble
SEND FUNCTION
4. You can hit Send and GPT will respond, Hit send a second time and GPT should continue the thought stream where it cut off from the same message (or)
5. You can hit Send a second time and if GPT has finished a thought or switched to a new Model - a New Chat Bubble will appear.
In a previous version, GPT did no. 4 and no. 5 inconsistently.
If the chat bubble was Greyed out - sometimes if I hit "Send", a new chat would continue but as part of the Greyed out bubble, but sometimes it would start a New Chat.
If Chat Bubble was an unfinished thought stream- Sometimes it would continue the thought where it was cut off, and other times it would respond to the same Prompt within the same Chat Bubble - repeating itself again.
If a Chat Bubble had a finished thought - or - if I switched models, hitting the send button again, would often cause a new Chat Bubble as expected.
It was often inconsistent - but it worked enough of the time, that you could gamble with it and get the results you wanted most of the time. One solution was instead of hitting send typing a random character (^). But I ONLY needed to do that if the chat bubble was left unfinished.
In this new version (with default settings) it no longer functions that way. Now The chat bubble never "continues" it's thought process (or) starts a new chat bubble when I hit send twice - hitting send ALWAYS continues the chat bubble with a new prompt or a repeat of the prompt - if the Chat bubble is Grey, an Unfinished Thought or Complete thought - it behaves the same. It continues the thought stream in the Chat bubble starting the Prompt over.
The only solution is to manually type characters instead of hitting the "^" button, or to manually tell it to continue where it left off - which uses up tokens. It seems with each new improvement, the send feature itself becomes a little bit more broken.
Questions:
Does the Send button work as intended?
Are there any plans on the developer's end to resolve the Send Feature? (or does it work the way dev's want it to)
In src/lib (or) Chat.svelte (?) or ChatRequest?? - what part of the code is controlling the logic for how the Send works?
The text was updated successfully, but these errors were encountered:
I don't know how to explain this, and as there are no version numbers to chatgpt-web, I cannot tell you what version I had before the update. I am not familiar with the terminology to properly express myself. I am going to do my best:
The Send Feature is buggy. I thought the update would improve the send feature, but it has only gotten worse for my usecase. I don't know if this is by design and is in fact better for everyone else, or just me who finds it a challenge.
Here are the basic features:
CONDITIONS
SEND FUNCTION
4. You can hit Send and GPT will respond, Hit send a second time and GPT should continue the thought stream where it cut off from the same message (or)
5. You can hit Send a second time and if GPT has finished a thought or switched to a new Model - a New Chat Bubble will appear.
In a previous version, GPT did no. 4 and no. 5 inconsistently.
It was often inconsistent - but it worked enough of the time, that you could gamble with it and get the results you wanted most of the time. One solution was instead of hitting send typing a random character (^). But I ONLY needed to do that if the chat bubble was left unfinished.
In this new version (with default settings) it no longer functions that way. Now The chat bubble never "continues" it's thought process (or) starts a new chat bubble when I hit send twice - hitting send ALWAYS continues the chat bubble with a new prompt or a repeat of the prompt - if the Chat bubble is Grey, an Unfinished Thought or Complete thought - it behaves the same. It continues the thought stream in the Chat bubble starting the Prompt over.
The only solution is to manually type characters instead of hitting the "^" button, or to manually tell it to continue where it left off - which uses up tokens. It seems with each new improvement, the send feature itself becomes a little bit more broken.
Questions:
The text was updated successfully, but these errors were encountered: