the game its constantly crash

I have been trying to look it up but I just can’t figure out how to word it correctly on google. But what I wanted to know is say you have a 131k character memory. When it reaches that limit does it completely forget those 131k characters? Or does it just forget the oldest interactions for new ones? I am just not sure how much 131k can really remember so I don’t know if I should keep reminding it of things so it does not forget.

Hello, how were you able to increase the speed of responsiveness and response generation from AI?

Is there any restrictions in game text prompt's length?

the sentences keep cutting off and when i try to limit the token output it keeps doing it. is there a way to fix that?

So two suggestions, the first would be a token counter or token percentage usage so the player can see how many tokens have been used by his current roleplay. Like it the player has an AI model that provides 10K tokens, if the player uses a hundred tokens, then he's used 1% of the available tokens. 

I am quite ignorant of AI and how the system works so please ignore if this is impossible but when the AI reaches a certain point, it could create a kind of save file or data packet with the relevant data done so far. As an example, the player has a make shift spear, is weak to hypnotism, has stats A, B and C etc in a formalised file. If the data has already been generated, would the token use be lowered to check the created data packet for items etc, physical characteristics, statistics etc rather than using tickets to regenerate more tokens? Or would it be the same? A dynamic inventory/state file might be more efficient means of token usage? Being cheeky, I asked the AI first. Here's the question and response, hopefully it might help:

The following is in the context of running a game generated entirely by ai.

Is it possible for when the AI reaches a certain point, it could create a kind of save file or data packet with the relevant data done so far. As an example, the player has a make shift spear, is weak to hypnotism, has stats A, B and C etc in a formalised file. The aim of this is to minimise the use of tokens and generate files that can be created to add themes, statistics etc to the game so a horror game doesn't find itself with comedy elements, a pornographic game maintains details so the player character doesn't start sucking cock in a glory hole only to get a follow up scene that they're in a bed in a model

Assistant mistral-nemo-instruct-2407Yes, it's possible for an AI like me to create save files or data packets containing relevant information about the game state, character attributes, and other pertinent data. This approach can help manage token usage more efficiently and maintain consistency in the generated content.

Here are some ways we could implement this:

{   "player": {     "name": "John Doe",     "weapon": "make-shift spear",     "resistances": {       "hypnotism": 0.3     },     "stats": {       "strength": 5,       "intelligence": 7,       "endurance": 6     }   },   "game_settings": {     "genre": "horror"   } } 

how would one go about gaining traits mid game instead at just the start?

Can i use a Entity to represent the player? 

Can I add armor as functional stat?

So armor could take part of the damage, be damaged and could be described. Could be restored by player and could be taken off.

Have you any advices?

Hi, I’m enjoying the game so far. I’ve got two questions/suggestions.

1. Would a Linux build be workable to allow switching the AI endpoint etc without needing to deal with Wine?

2. Would integrating a scripting language (Lua, Javascript, or whatever), to force sending a prompt, manipulating stats in a certain way, or triggering a gameover screen under certain conditions, as part of world construction be possible? More complex concepts exceed the bounds of local AI pretty easily and the AI isn’t always reliable for clear cut rules. Offloading hard rules in world creation should provide some possibilities, for example, your stomach and satiety hit zero triggers a gameover by starving, your stamina hits zero triggering your character collapsing from exhaustion, a specific character is present when you enter a particular room for the first time, etc.

After some experimenting I have found that editing the first line in the "Game Text Prompt" in the System Prompts section of the settings has provided me with a more enjoyable experience. I replace the section saying "direct the player" with a phrase that contains "provide the player with a narrative decision point" and then something about giving the player the ability to move the story forward or give the player options to continue the narrative. The second half of the phrase has been the variable part of my experiments. For me it has curbed the habit of the AI producing an entire short story with a conclusion (with the one paragraph limit turned OFF). I have encountered some instances of the AI putting lists of possible actions to take in the event text instead of in the choice boxes, but it stops if I start a new game. It's not perfect but something I have been trying.

Hello! I was wodnering if there is possibility for ai to create and modify json files. Or something similar to keep track and reminditself of transfromations, npc. locations and apperance of them. I am trying in various ways but a im not sure if i reinforce anything in that way.

Will you ever consider larger sizes or new body parts (i.e. futa dicks)

Here's the enormous choices I received when I touched the [Choices Prompt]s a bit... I think I need to tweak it a bit more, It's basically serving everything to me on a silver platter, the choices are great but altering them is a hassle with these crazy things.

Would you ever consider adding fields to control the parameters ( like temp, top_p, etc ) for the model for hosted versions like OpenRouter and DeepInfra as well as local hosted solutions? Looking through the docs, this appears to be part of the base OpenAI API specs.

{     
  "model":"shuyuej/Mistral-Nemo-Instruct-2407-GPTQ",
  "temperature": "some temp",
  "top_p": "etc",
  "min_p": "etc",
  "messages":[         ...         ],     
"max_tokens":1024,"stream":true,"stop":["\n"] }

https://platform.openai.com/docs/api-reference/completions/create
https://openrouter.ai/docs/api-reference/overview
https://lmstudio.ai/docs/app/api/endpoints/openai
( Deep Infra API docs require going to deployment to view otherwise I would list them here too)

I was able to replicate the POST requests your game makes with some added fields and it didn't immediately fail though the output was interesting.
I'm not sure what effects this would have on Oobabooga but I would hope it wouldn't break anything if its following the API specs

Asking since DeepInfra and OpenRouter AI only seem to support setting these through the API and some models might benefit from fine tuning these parameters.

I would like the input box to be focused when finished generating text, sometimes I type text directly without selecting a answer and it's troublesome when it's not focused.