I was very skeptical at first, but after looking at the Kickstarter and Discord, this can easily be the VIVA successor we needed. Thank you for the hard work and I wish you much luck.
What would be interesting if there was an android/windows/linux/mac versions with apis already included so inexperienced people could explore this game in its entirety. looks great btw.
Sounds like you are using too strong of an AI file for your hardware or settings aren't set quite right. I had the same issue but It took me some fiddling on settings before I got it to work. When Kobold_cpp first starts, it is looking at your CPU. Mine is not great so I had to switch it to use my GPU more through I think it was Vulkan Old CPU.
this is very promising and i myself have been testing and messing with connecting LLM's to a game world. I am wondering about though, when using local LLM does the game tell the LLM about the robots stats, body and what it can perform in terms of actions? i read that currently there is a limitation with function calling but would love to hear more about how this works :)
do you parse the actions from the text generated, or perhaps make the llm generate both speech and actions seperately or something like that?
← Return to game
Comments
Log in with itch.io to leave a comment.
Please add support for Portuguese (Brazil) language.
I am wondering, will this game come on steam and will it have a DLC option to get the AI that way?
Please make one on mobile🥺🥺
when is it coming for mac?
[Context Shifting: Erased 7 tokens at position 2]
Processing Prompt (10 / 10 tokens)
Generating (77 / 512 tokens)
(EOS token triggered! ID:128009)
[10:06:31] CtxLimit:3661/4096, Amt:77/512, Init:0.05s, Process:1.29s (129.4ms/T = 7.73T/s), Generate:14.34s (186.2ms/T = 5.37T/s), Total:15.63s (4.93T/s)
Output: <Generated Message>
Generate: The response could not be sent, maybe connection was terminated?
the connection getting terminated. i run local LLM with kobaldcpp as per the documentation of this game.
can you able to analyze what's going on here?
If it took you 14 seconds to generate a response perhaps your computer isnt powerful enough to generate offline? What are your computer specs?
I don't have GPU. I run Llama 3B parameter model. it runs in CPU only hence response is slower . 12 core CPU.
I use Wine compatible layer to run your game in Linux.
Thanks for replying and trying to help. I like your game testing it. i will buy it soon
Hi sgthale! Apparently, the game isn't labeled as compatible on itch.io, preventing installations. please fix I rlly wanna play!
You can manually download it and install it just now with the itchio client.
Help! what should I do if I run this application and it returns error 0x00000050?(I'm on windows 7)
i dont know if it supports windows 7? Maybe you need a better graphics card with DX12
Will being able to run through an LLM be an option?
of course you can already do that!
Oh, wow, thats cool!
I was very skeptical at first, but after looking at the Kickstarter and Discord, this can easily be the VIVA successor we needed. Thank you for the hard work and I wish you much luck.
thanks
Please add Russian language support. A very large audience needs it
okay!
What would be interesting if there was an android/windows/linux/mac versions with apis already included so inexperienced people could explore this game in its entirety. looks great btw.
pleaseeeee add modding support
Mod support is on the Kickstarter goal list above. So, yes, there will be Steam Workshop mod support.
How do you play it? I am on a fire tablet and it only goes to my Amazon files
little bro get of itch and go play with playdough
Can I use Gemini AI's api key?
I get an issue using Kobold_CPP, where I talk, it sends, and it generates fine, but the game connection seems to terminate after generation.
im also with the same struggle, lemme know if you get any responses plss
That's what It's doing here, though terminating before generation, saying it can't connect, even though it is.
Sounds like you are using too strong of an AI file for your hardware or settings aren't set quite right. I had the same issue but It took me some fiddling on settings before I got it to work. When Kobold_cpp first starts, it is looking at your CPU. Mine is not great so I had to switch it to use my GPU more through I think it was Vulkan Old CPU.
I think it would be great if they released a free beta on Steam. 👀
yeah we are not at that stage yet
this is very promising and i myself have been testing and messing with connecting LLM's to a game world. I am wondering about though, when using local LLM does the game tell the LLM about the robots stats, body and what it can perform in terms of actions? i read that currently there is a limitation with function calling but would love to hear more about how this works :)
do you parse the actions from the text generated, or perhaps make the llm generate both speech and actions seperately or something like that?
yeah I inject messages to the chat for the game.
yes I parse actions manually sometimes depending on the action.
So i kind of took a look at the code and if i am not mistaken, are you not sending the tool list when using a local llm?
are you sending the tools as
messages=[{"role": "user", "content": "What is the weather like in Paris today?"}], tools=toolsor is the tools like this?
messages=[{"role": "user", "content": "What is the weather like in Paris today?", "role": "tools", "content":"{{tool_list}}"}]neither, loop up tool calls on OpenAI API docs
what a fascinating project, I'm eager to try it!