in a new chat please. this one is dead 
ok, you did not answer this question from above. lets do this first
update: sry didnt see, you did.
==> Safari is not working. Use Chrome.
I had mentioned above, I am on safari on a Mac.
got it. Let me check on chrome.
1 Like
it works and gives output - but every time I get this error There was an error processing your request: <html> <head><title>413 Request Entity Too Large</title></head> <body> <center><h1>413 Request Entity Too Large</h1></center> <hr><center>nginx/1.26.0 (Ubuntu)</center> </body> </html> <!-- a padding to disable MSIE and Chrome friendly error page --> <!-- a padding to disable MSIE and Chrome friendly error page --> <!-- a padding to disable MSIE and Chrome friendly error page --> <!-- a padding to disable MSIE and Chrome friendly error page --> <!-- a padding to disable MSIE and Chrome friendly error page --> <!-- a padding to disable MSIE and Chrome friendly error page -->
@jacklearner on which project / prompt? Should not happen with the tic tac toe game from before. For bigger project it can be that it is to big to fit in in the context size of the llm.
Can you provide the chat export of what you are trying please?
This is the chat I put up → “Sign Up and Sign In Page, Forgot Password option are missing” to which it replied with the error message.
yeah but on what codebase? I dont think thats your initial prompt?
so all chat messages combined are considered for context input? My overall chat is definitely huge.
Yes, otherwise the LLM can not know what to implement if it does not have the context.
Make sure in Settings->Features, Context Optimization is enabled. But if this is already, then its over all just to big I guess, at least for the LLM you working with.
1 Like
Appreciate your help here and getting some good results with Bolt!
I am also seeing this behavior where at some point deep into a fairly lengthy conversation bolt stops modifying files / running terminal commands.
The same model on a new project from scratch creates / modifies files and executes commands to run the project fine but on the project with long context it will not do this anymore.
Context optimization is enabled.
It does not appear to be a model issue as the model continues to output in the same format it was previously but Bolt isn’t taking those artifacts and attempting to apply them it seems.
Model server in this case is ollama and models qwq:32b-q8_0 / qwen2.5-coder:32b-instruct-q6k all produce the same behavior outlined above.
Afaik ollama behavior is to silently discard context over the model default which presumably degrades the responses as the project evolves past the context window but the output format should remain the same so unclear why changes at some point stop being applied.
Can provide additional details as needed or open an issue if that’s a better place to further this discussion.
@osc123 thanks. I saw this sometimes also. If you got details, reproduceable maybe as well, yes open a github issue with it.
No problem at all! Issue 1502 has been opened here
1 Like
@leex279 I checked with other models and for every model now I get blank screen with 3 dots. This is stuck here since past 5 mins, the prompt is hardly of 5 lines.
@jacklearner strange. do you have any errors in terminal log or the dev-console? must be I think.
{
“name”: “TypeError”,
“message”: “Failed to fetch”,
“stack”: “TypeError: Failed to fetch\n at A4e (https://mydomain.com/assets/Header-Cju3Om0T.js:126:4134)\n at https://mydomain.com/assets/Header-Cju3Om0T.js:1107:12154\n at https://mydomain.com/assets/Header-Cju3Om0T.js:1107:13482\n at le (https://mydomain.com/assets/Header-Cju3Om0T.js:1221:33)”,
“component”: “Chat”,
“action”: “request”,
“error”: “Failed to fetch”
Looks like you are not on a local setup. I think its better you open a seperate topic for your case and describe your exact setup there first, with screenshot also from the system/terminal and the errors.
1 Like