Others Formamorph [v1.1.5] [FieryLion]

[Serpream]

So my last questions before next version @fierylion -
I take it the Status prompt it meant to be the place you want any updates to take place and be invisible to the printed event/actions?

I keep getting odd situations where the Event is extremely clear a trait should be applied, but the stats prompt adds something different or not at all.

I am going to do my final pass today and just wait for next app version (as been stuck hitting weird limits).

Final question is if you will be adding a local Database to app for AI to dump stuff back into? Or will it be like current Traits and only there if manually added and referenced?

I am tempted to setup a local MySQL database and experiment for table storage and consistancy to free AI memory.

 

[Serpream]

omgomg omg I AM FINALLY MAKING HUGE PROGRESS!
Main event questions are now inline with the action prompts. Stats working sensibly. Continuity respected.
Just got to fix a character loop issue when too much dialogue and can release V5 >.<

 

[Nouvi]

omgomg omg I AM FINALLY MAKING HUGE PROGRESS!
Main event questions are now inline with the action prompts. Stats working sensibly. Continuity respected.
Just got to fix a character loop issue when too much dialogue and can release V5 >.<

That's awesome. Can't wait to play it once you release it. Continuity have by far been the biggest issue so far, but having that fixed... the potential to keep good scenes going is gonna be insane. (well, aside from the part everything is asking for consent for absolutely every action on a retarted level... until I fixed that on my own, at least for my own use)

 

[Deathend]

some question i have? is it or will be posible to use an ia capable of drawing? having the ia narrate the story and draw it on player prompt whould be awesome.

 

[Deathend]

today, traits system started working for me again, no idea why. the only change i did was changing from ollama to Lmstudio and change to the Q5 mistral model. with should not have any effect on the trait system working or not.

Also, can we have a force to stop button? the amount of times the Ai went too out of its way and i had to wait for it to finish just so i could rollback is insane.

 

[Gizza]

I have been attempting to figure out the Trait system myself but have had little success.
Can traits that you create in the traits section, be acquired through in game events rather then selecting them at the start of the game?
If so, what should the prompt entail to trigger it?

 

[Nouvi]

I have been attempting to figure out the Trait system myself but have had little success.
Can traits that you create in the traits section, be acquired through in game events rather then selecting them at the start of the game?
If so, what should the prompt entail to trigger it?

To my experience, traits cannot actually be gained outside of character customization unfortunately. There might be a way to do it through setting up rules for it in the right places, but I have no idea how complex that would be, and I don't think anyone has experimented with it yet.

 

[Serpream]

To my experience, traits cannot actually be gained outside of character customization unfortunately. There might be a way to do it through setting up rules for it in the right places, but I have no idea how complex that would be, and I don't think anyone has experimented with it yet.

Oh boi guess you did not read my early posts and proof of concepts... Got day free due to Food poisoning and fixed LM Studio crashing with update, so will get my last draft out today with stuff in it.

 

[Nouvi]

Oh boi guess you did not read my early posts and proof of concepts...

Oh I have, but I was assuming he was talking in regards to the actual traits system in the game, not the system that you've managed to create that largely functions separately from it. (Which is damn amazing btw)... unless you actually DID get it to interact with it rather than through tables.

On a side note... AAAAA WHEN WILL YOU RELEASE YOUR NEW UPDATE! This day's gonna be a painful wait...
That being said, been messing some with custom traits using your world while waiting, to actually some quite amazing results.

Spoiler: An example

You don't have permission to view the spoiler content. Log in or register now.

 

[WishingForAnotherName]

I've been trying out a non-abliterated model, the Mistral-Nemo-Instruct-2407 Q4_K_S. So far it's an improvement over the abliterated version. On the other hand, it seems like it's even more overly long-winded.

Seriously, I don't mind reading, but the sheer length of the replies and the extent to which the AI goes beyond the scope of your prompts can be frustrating.

 

[Deathend]

I've been trying out a non-abliterated model, the Mistral-Nemo-Instruct-2407 Q4_K_S. So far it's an improvement over the abliterated version. On the other hand, it seems like it's even more overly long-winded.

Seriously, I don't mind reading, but the sheer length of the replies and the extent to which the AI goes beyond the scope of your prompts can be frustrating.

try to put a rule in the narration setting section limiting the paragraphs to the amount you want, and if it does not obey that rule put it on every prompt you give in the game, that worked for me

 

[WishingForAnotherName]

I've got to try something. Any tips on how I should word it or format it?

So far I haven't messed with anything beyond trying new models in LMStudio and fiddling with the Max Memory and Max Output Tokens settings.

 

[Serpream]

Oh I have, but I was assuming he was talking in regards to the actual traits system in the game, not the system that you've managed to create that largely functions separately from it. (Which is damn amazing btw)... unless you actually DID get it to interact with it rather than through tables.

On a side note... AAAAA WHEN WILL YOU RELEASE YOUR NEW UPDATE! This day's gonna be a painful wait...
That being said, been messing some with custom traits using your world while waiting, to actually some quite amazing results.

Spoiler: An example

You don't have permission to view the spoiler content. Log in or register now.

Glad your excited. Sadly despite being sick from food poisoning. I still got called out to 2 meetings and a database fix on a Sunday!!
I also just tried to do a quick release but created a logic loop somewhere. Here is hoping to free time tomorrow >.<

*And just fixed it. its 23:23 and I need sleep and still have all Player logic to finish, then will release. Sadly breached way past the 4096 token limit. So curious if a more efficient way to set these rules...*

 

[Nouvi]

breached way past the 4096 token limit.

I don't think that's too big of an issue. Increasing that a bit in LMStudio is easy, and doesn't affect much unless you really go overboard. (e.g., doubling it is hardly noticeable, but doing like 5x or 10x the default would likely make it a little slow to respond)

 

[BackFatMan]

Hey if anyone could, could you make a pokemon theme world in not smart enough and would just copy and paste shit from wiki

 

[Breedking]

Sorry about the lack of updates on my end, work has been destroying me. For anyone wanting to try new AI, (I'm getting constant user permission issues with llama based AI) qwen2.5-7b-instruct-uncensored is extremely fast. I'm just trying it out, s so no review yet. Normally I've been using the previously linked abliterated AI.

Edit: no go, even at 20000 contex, it rapidly loses track of what is going on.

Anyone know of a fix for the llama Ai for LM studio?

 

[GucciNinja]

(me after reading the entire thread)

 

[Gizza]

Glad your excited. Sadly despite being sick from food poisoning. I still got called out to 2 meetings and a database fix on a Sunday!!
I also just tried to do a quick release but created a logic loop somewhere. Here is hoping to free time tomorrow >.<

*And just fixed it. its 23:23 and I need sleep and still have all Player logic to finish, then will release. Sadly breached way past the 4096 token limit. So curious if a more efficient way to set these rules...*

How are you seeing your over the token limit out of curiosity?

I have been adjusting the Game text prompt's as its become increasingly irritating the level of hand holding the AI requires some times. Despite explicitly giving the AI the permission to proceed with action A or to do action B etc, it instead seems to reapt itself in some instances and then asks for permission again.
I suspect after adjusting it ill find myself having the opposite problem of the AI running wild and having to reign it in again, guess we'll find out.

 

[Nouvi]

How are you seeing your over the token limit out of curiosity?

I have been adjusting the Game text prompt's as its become increasingly irritating the level of hand holding the AI requires some times. Despite explicitly giving the AI the permission to proceed with action A or to do action B etc, it instead seems to reapt itself in some instances and then asks for permission again.
I suspect after adjusting it ill find myself having the opposite problem of the AI running wild and having to reign it in again, guess we'll find out.

If you're over the token limit, LM Studio's developer logs will throw an orange error telling you so, and by exactly how much.

Also, changing the limit is done by editing the AI model's config under My models -> Actions, (the cogwheel there) and changing the Context Length value. It does warn about possible stability issues, but I have not really noticed anything after increasing it to 10k, and most models support 128k.

 

[Serpream]

How are you seeing your over the token limit out of curiosity?

I have been adjusting the Game text prompt's as its become increasingly irritating the level of hand holding the AI requires some times. Despite explicitly giving the AI the permission to proceed with action A or to do action B etc, it instead seems to reapt itself in some instances and then asks for permission again.
I suspect after adjusting it ill find myself having the opposite problem of the AI running wild and having to reign it in again, guess we'll find out.

LM Studio will straight out allow you debug logs and can see for example my current version costs 4192 tokens, which is over the 4096 and will not work unless I extend the tokens, which I do up to 8000.

As for getting stuff finished. Been insanely ill and busy but might have free time Sunday. I have all the systems working great, but the continuity issues and then looping back into Events still happens so randomly it became infuriating to try and fix. So taken a break to refresh to see if I spot anything obvious on return.