- Jan 18, 2019
- 223
- 417
ya it died for a few hoursdid the ai server or so die i cant start any games and dam i was mid story 2. browser verison
ya it died for a few hoursdid the ai server or so die i cant start any games and dam i was mid story 2. browser verison
actuly i havent messed with it and sorry if this was ask is there a way to save your progress in browser mode and come back it itya it died for a few hours
yeah there's a save and load option in settingsactuly i havent messed with it and sorry if this was ask is there a way to save your progress in browser mode and come back it it
Fixed in v1.1.5 hotfixgot it on the website, just added g then turn off 3d character.Just want to chime in that I am also getting this error message. It will initially let me edit and save any world no problem. However, when I go in and try to edit and save a world I have already edited I get the error message. This is happening on both the browser(using chrome) and downloaded versions.
Glad to see all my previous advice was spot on then with 2 weeks experienceAs a former programmer with 3 years of experience myself, I can recommend a couple of tips that helped me to get better responses from the AI (although it is not necessary to have programming experience, the prompts can be optimized if you understand how they work, information that I happen to have):
Have fun y'all <3
- Modify the system prompts in the settings, the default prompts are repetitive and can lead to confusion on the part of the AI, especially with the model it includes (shuyuej/Mistral-Nemo-Instruct-2407-GPTQ). Try things like "Act as an immersive game narrator in a dark fantasy world. Describe scenes vividly, but avoid modern or out-of-character knowledge."
- Try to use safer and, if possible, local models.
- Avoid endpoint URL's like "lyonade dot net" as they are unstable and can experience crashes as I have experienced.
- If you install local AI models, make sure you have suitable hardware, it is not always necessary to have a NAZA PC to run an AI, look into models that are low in RAM and VRAM consumption.
- Establish clear parameters.
- Install uncensored AI's, there are several models available on sites such as Hugging Face. Always be careful and don't play with commands you don't know.
if you force stop the generation. you can't rollback sometimes. good thing you can delete texts so the AI doesn't use that page.Fixed in v1.1.5 hotfix
you can use tab then go through each option until you land on the menu button then use enter or space. or do ctrl plus scroll wheel to zoom out.for the browser version is there a way to get ride of the 3 links on right side of the page they cover up the setting button
bottom right theres a button to fullscreenfor the browser version is there a way to get ride of the 3 links on right side of the page they cover up the setting button
ya was trying not to use fulle screen as multitasking while the ai loads its reponses but second question how does it track time. it like if we metnion it will keep track or is just a place holder for now aboue the stats and stuffbottom right theres a button to fullscreen
I used to combine the AI prompts and let the AI track time after each event, but since v1.1, to speed things up by 10 times, I split everything into 3 different AI calls. Since I can't waste an entire AI call just to track time I've disabled it for now, so right now every action takes 1 hour exactly. I'm thinking of adding optional toggles for the AI to track more things such as time passed, keep notes of new entities encountered, adding/removing traits and such.ya was trying not to use fulle screen as multitasking while the ai loads its reponses but second question how does it track time. it like if we metnion it will keep track or is just a place holder for now aboue the stats and stuff
ah ya deffently need a bit of tweekend had the ai restart the story itself to the beging and the only fix was rollback to a few convos agoI used to combine the AI prompts and let the AI track time after each event, but since v1.1, to speed things up by 10 times, I split everything into 3 different AI calls. Since I can't waste an entire AI call just to track time I've disabled it for now, so right now every action takes 1 hour exactly. I'm thinking of adding optional toggles for the AI to track more things such as time passed, keep notes of new entities encountered, adding/removing traits and such.
103b model is just too big, for only 22gb vram you can go for gguf of 24b model maybeThere's a new version posted on itch.io that might speed the game up and has some bug fixes.
I'm trying the mradermacher/Rogue-Rose-103b-v0.2-i1-GGUF/Rogue-Rose-103b-v0.2.i1-IQ1_S.gguf AI (22gb so a lot of computers can't run it) and will report back. Supposedly it's good for narration.
Edit: It utterly nuked my GPU and I'm running a 4070 Super. No go for me.
I didn't really read your posts but I'm glad it was helpful.Glad to see all my previous advice was spot on then with 2 weeks experience
I am finding tables are working wonders for the AI memory issue so far.
*Will not have time to finish up now until Thursday as waited on Hotfix![]()