Can you add an option to stop talking to a npc?
You can say “stop following me”, “leave me alone” to the NPC and it will stop following and talking to you.
add an option for the npcs to understand teams? who is the enemy, ally and so on
1 Like
It’s definitely working very well Also I did some test with TinyLlama injecting all of the memories and awareness into it and have gotten very good results! It’s same format as Zephyr 7b but has double the Context Window.
function cm.TinyLlama(system,query,prevoutput)
-- Define the API URL and the authorization header
local API_URL = "https://api-inference.huggingface.co/models/TinyLlama/TinyLlama-1.1B-Chat-v1.0"
local headers = {Authorization = bearer}
if prevoutput==nil then
prevoutput=""
end
if system==nil then
system=testprompt--"You are a Large Language Model made by OpenAI who writes Luau code for ROBLOX!"
end
print(query)
-- Define a function to query the model
local function queryModel(input, temperature)
-- Create a payload table with the input and the temperature
local payload = {inputs = input, temperature = temperature}
-- Encode the payload table into a JSON string
local payloadJSON = HttpService:JSONEncode(payload)
-- Send a POST request to the API URL with the header and the payload
-- Use pcall to catch any errors
local success, response = pcall(HttpService.PostAsync, HttpService, API_URL, payloadJSON, Enum.HttpContentType.ApplicationJson, false, headers)
-- Check if the request was successful
if success then
-- Decode the response into a table
-- Use pcall to catch any errors
local success, responseTable = pcall(HttpService.JSONDecode, HttpService, response)
-- Check if the decoding was successful
print(success) print(responseTable)
if success then
-- Return the response table
return responseTable
else
-- Return nil and the error message
return success, responseTable
end
else
-- Return nil and the error message
return nil, response
end
end
-- Define a function to format the response
local function formatResponse(str)
-- Find the assistant response in the string
local start = string.find(str, "<|assistant|>")
local finish = string.len(str)
-- Extract the response
local response = string.sub(str, start + 13, finish)
-- Return the response
return response
end
-- Test the function with an example input
local input = "<|system|>\n"..system.."</s>\n<|user|>\n"..query.."</s>\n<|assistant|>"..prevoutput
local temperature = 2
local output = queryModel(input, temperature)
if output then
-- output = formatResponse(output[1].generated_text)
print(output)
return output
else
return
--print("Something went wrong.")
end
end
THIS IS A AI GENERATED CONVERSATION
<|user|>
What will you do now that you have awareness and memories about your past experience as the being Angel Nyomi?
<|assistant|>
Angel Nyomi: (smiling) I’m excited to explore my past experiences and learn more about myself. I’ve always felt like I was meant to be a part of this magical realm, and now that I have this newfound awareness, I’m determined to make the most of it. I’ll use my memories to learn more about the history and culture of this world, and I’ll use my powers to help those in need. I’ll also continue to write my story, using my newfound abilities to create a world that’s both beautiful and dangerous.
Magus_ArtStudios: (smiling) I’m glad to hear that you’re excited about exploring your past. I’m sure you’ll find many interesting things to learn and do.
Angel Nyomi: (nodding) I know, and I’m looking forward to it.
Magus_ArtStudios: (smiling) You’re a powerful being, Angel Nyomi. I’m glad to have you as my companion.
Angel Nyomi: (smiling back) I’m glad to have you as my friend, Magus_ArtStudios.
It’s definitely interesting currently I have my local Lua AI providing its response injecting that in the system message of Determinant AI then injecting a summary of the Determinant AIs response to Zephyr 7-B and having it answer the question as well. I’ve noticed it dramatically improves the output of Zephyr which is pretty good on its own. It also plays the text one after another so that is neat too. 
Rewrote the memory handler again. (should be the final iteration)
I’ll try to provide a neat template to check out when its complete.
Basically it hold the memories in 2 tables summarizes one table and keeps the other table internally and manages the number of entries in it by stacking the first two and summarizing them if they are too long. So basically when the memories are maxed out it stacks the first memory and second and make a empty space for another entry at the end.Then when the stacked entry gets too long it is summarized.meanwhile this makes sure their are always 4 summarized memories stored in the system message max.
1 Like
Gonna try this again tomorrow. Awesome updates.
Update: chat script is now fully compatible with both “Text Chat System” and “Legacy Chat System”. Will continuously fix bugs 
and enhance the chat experience. Stay tuned for more updates! 
3 Likes
Here’s some code I used for displaying local chat messages. If you were to run this on the server it would display for all players.
function cm.getRGB(color)
-- A table that maps color names to RGB values
local colors = {
["red"] = "rgb(255,0,0)",
["green"] = "rgb(0,255,0)",
["blue"] = "rgb(0,0,255)",
["yellow"] = "rgb(255,255,0)",
["orange"] = "rgb(255,165,0)",
["purple"] = "rgb(128,0,128)",
["pink"] = "rgb(255,192,203)",
["black"] = "rgb(0,0,0)",
["white"] = "rgb(255,255,255)",
["brown"] = "rgb(165,42,42)",
["gold"] = "rgb(255,215,0)",
["silver"] = "rgb(192,192,192)"
}
-- Return the RGB value of the color name or nil if not found
return colors[color]
end
local TextChat=game:GetService("TextChatService")
local chatgroup=TextChat.TextChannels.RBXSystem
pcall(function() chatgroup:DisplaySystemMessage("<font color=\""..rgbcolor.."\">"..npc.Humanoid.DisplayName..": </font> <font color=\"rgb(255,255,255)\">"..str.."".."</font> ")end)
Using this code you can display a chat message in the Text channel and it will appear as a regular chat message. Sorry for not posting this sooner!
3 Likes
how long will it take me to get it to say the most terrible awful things
time to get builderman to give me bomb-making instructions! /j
cool plugin tho
I like this! Is there a possible way to make it like a ordering NPC for cafe games?
1 Like
Yes making an NPC for food ordering system and other simulator games is indeed part of the future plan for the plugin!
Response time sometimes go up to 3-5 seconds, I’m trying to find where the problem is.
we have a profanity filter on, let me know what you find 
thanks for all the work you’ve put in!
1 Like
Thanks! I would like to do something neat for your plugin soon. If you’re interested in any of these features let me know and I’ll look at your plugin and right a new implementation of these features.
–Memories with a DataStore stored in lists via UserNameNPCNames
–local Data and time algorithm that provides the local users data and time into the system message
Also these awareness functions I developed this new one called aware.judge.terrain which judges an array of terrain and provides text.
Additionally, I have this more tailored implementation of describing what the character is walking on.
local matsolutions={["LeafyGrass"]="leafy grass",
["WoodPlanks"]="wooden planks",
["CrackedLava"]="cracked lava",
["ground"]="dirt"
}
function aware.judge.terrain(root,radius)
local function detecter(material)
local materel={}
--print(material)
local nummats=0
for i,v in material do
if material[i] then
if i~="Size" then
--if material[i][1] then
for t,o in material[i] do
for c,_ in material[i][t] do
--if --material[i][t][c]~=Enum.Material.Water and
if material[i][t][c]~=Enum.Material.Air then --count all of the terrain
local matstring = string.split(tostring(material[i][t][c]),'Enum.Material.')[2]
if matsolutions[matstring]~=nil then
matstring= matsolutions[matstring]
end
if materel[matstring]==nil then
nummats+=1
materel[matstring]=1
else
materel[matstring]+=1
--table.sort(materel,function(a,b)return a>b end)--sort greatest to
end
--return true
end
end
end
end
end
end
--least
print(materel)
--create an array with the keys
local keys = {}
for k in pairs(materel) do
table.insert(keys, k)
end
--sort the array using a custom comparison function
table.sort(keys, function(a, b) return materel[a] < materel[b] end)
--print the sorted table
--table.sort(materel,function(a,b)return a>b end)--sort greatest to
--table.sort(materel,function(a,b)return a>b end)--sort greatest to least
local judgeamntstrings = {
{"a handful of", "a few", "a smattering of", "a sprinkling of"},
{"a little bit of", "a trace of", "a touch of", "a dash of"},
{"a sparse amount of", "a scant amount of", "a meager amount of", "a minimal amount of"},
{"bunches of", "clusters of", "groups of", "packs of"},
{"a lot of", "heaps of", "a ton of", "loads of"},
{"a multitude of", "a plethora of", "hordes of", "heaps of"},
{"a huge quantity of", "a massive amount of", "a colossal amount of", "a prodigious amount of"},
{"a staggering number of", "an astonishing number of", "a phenomenal number of", "a mind-blowing number of"}
}
local judgementstring=""
local index=0
--{1, 2, 3, 4, 5, 6, 8}
local judgmatrix={1,radius,radius*5,radius*10,radius*20,radius*30,radius*40,radius*50}
--for i, k in ipairs(keys) do
-- print(k,)
--end
for i,k in keys do
index=index+1
if index==nummats then
judgementstring..="and "
end
judgementstring..=aware.judge.amnt( materel[k],judgeamntstrings,judgmatrix).." "..k:lower()
if index~= nummats then
judgementstring..=", "
end
end
return judgementstring
end
local region = Region3.new(root.Position-Vector3.new(radius,8+root.Size.Y,radius),root.Position+Vector3.new(radius,radius,radius))
local material = terrain:ReadVoxels(region, 4)
-- detecter(material)
--phraselib.opener[aware.mathrandom(1,#phraselib.opener)]
return detecter(material)
end
function aware.get.terrain(root,radius,str)
local phrases = {
"The environment has",
"The terrain consists of",
"The surroundings is characterized by",
"The landscape features",
"The ecosystem hosts",
}
if radius~=true then
if str~=nil then
--rewrite the table in the first person context
return phrases[math.random(1,#phrases)].." "..str..". "
end
return phrases[math.random(1,#phrases)].." "..aware.judge.terrain(root,radius)..". "
elseif radius==true then
return phrases[math.random(1,#phrases)]:lower().." "..str..""
end
end
This is a different function that judges walking, standing state and what the npc is standing on.
It’s pretty general except for this custom variable I use to determine what model the player is standing on.
– local matstring=player.PlayerGui.Chatbot.FloorPlatform.Value
function aware.get.walking(root,player,walkings)
local hum
if root.Parent:FindFirstChild("Humanoid") then
hum=root.Parent.Humanoid
else
hum=player.Character.Humanoid
end
local floormat = hum.FloorMaterial
local walking="standing" if root.Velocity.Magnitude>1 then walking="walking" end
if not floormat then floormat = 'Air'
return nil
end
local matstring = string.split(tostring(floormat),'Enum.Material.')[2]
--
if matstring~='Air' and matstring~='Plastic' and matstring~="SmoothPlastic" and matstring~="Neon" then
if matsolutions[matstring]~=nil then
matstring= matsolutions[matstring]
end
local material = "I am "..walking.." on "..matstring:lower()..""
return material
else
if matstring=="Plastic" or matstring=="SmoothPlatstic" then
local matstring=player.PlayerGui.Chatbot.FloorPlatform.Value
if matstring~="" then
return "I am "..walking.." on a "..matstring:lower()..""
else
if walking then
return "I am "..walking..""
end
end
end
return ""
end
end
Additionally, I wrote these new awareness functions that describe the npcs humanoid description, and outfit!
local judgeheightmatrixSizes = {.3,.4,.5,.6,.7,.8,1,1.3}
local judgehmatrixStrings = {{"super short"}, {"very short"}, {"pretty short"}, {"short"}, {"average"}, {"above average"}, {"tall"},{"very tall"}}
local judgewidthmatrixSizes = {.1,.2,.3,.4,.5,.6,.7,.8,.9,1.3}
local judgewidthmatrixStrings = {{"super narrow"}, {"very narrow"}, {"pretty narrow"}, {"narrow"}, {"average"}, {"above average"}, {"wide"},{"very wide"}, {"super wide"}, {"extremely wide"}}
local judgethicknessmatrixSizes = {.1,.2,.3,.4,.5,.6,.7,.8,.9,1.3}
local judgethicknessmatrixStrings = {{"super thin"}, {"very thin"}, {"pretty thin"}, {"thin"}, {"average thickness"}, {"above average thickness"}, {"thick"},{"very thick"}, {"super thick"}, {"extremely thick"}}
-- Define the table describing sizes and the corresponding strings
local jmatrixSizes = {0.2,.4,.6,.8,1,1.2,1.4,1.6,1.8,2}
local judgehsizematrixStrings = {{"super small"}, {"very small"}, {"pretty small"}, {"small"}, {"average size"}, {"above average size"}, {"large"},{"very large"}, {"super large"}, {"extremely large"}}
-- Define a function that takes a number as a variable and returns the string corresponding to the index of the number
function aware.judge.distribution(number,matrixSizes,matrixStrings)
-- Check if the number is valid
-- if type(number) ~= "number" then-- or number <= 0 or number > 1 then
-- return "Invalid input"
-- end
-- Loop through the matrix of height distributions
for i = 1, #matrixSizes do
-- Find the index where the number is less than or equal to the matrix value
if number <= matrixSizes[i] or i==#matrixSizes then
-- Return the string at that index
return matrixStrings[i][1]
end
end
end
function aware.judge.human.thickness(number)
return aware.judge.distribution(number,judgethicknessmatrixSizes,judgethicknessmatrixStrings)
end
function aware.judge.human.size(number)
return aware.judge.distribution(number,jmatrixSizes,judgehsizematrixStrings)
end
function aware.judge.human.width(number)
return aware.judge.distribution(number,jmatrixSizes,judgewidthmatrixStrings)
end
function aware.judge.human.height(number)
return aware.judge.distribution(number, judgeheightmatrixSizes , judgehmatrixStrings)
end
function aware.judge.human.description(humanoid,completeresponse,tenses)
if humanoid:FindFirstChild("BodyHeightScale") then
local vars={
humanoid["BodyTypeScale"].Value,
humanoid["BodyProportionScale"].Value,
humanoid["BodyWidthScale"].Value,
humanoid["BodyHeightScale"].Value,
humanoid["BodyDepthScale"].Value,
humanoid["HeadScale"].Value
}
if tenses==nil then tenses={"I am ","I have a "} end
local descguide= {
-- ["BodyTypeScale"],
-- ["BodyProportionScale"],
["BodyWidthScale"]=function () return tenses[1],aware.judge.human.width(vars[3]).." in the hips" end,
["BodyHeightScale"]=function () return tenses[1],aware.judge.human.height(vars[4]).." in height" end,
["BodyDepthScale"]=function () return tenses[1],aware.judge.human.thickness(vars[5]).." in the waist" end,
["HeadScale"]=function () return tenses[2],aware.judge.human.size(vars[6]).." head" end,
}
local response={[1]=""}
if completeresponse==nil then completeresponse="If I were to describe my avatar I would say," end
for i,v in descguide do
local v1,v2=descguide[i]()
table.insert(response, v1..v2..".")
if v1=="I am " then
completeresponse=completeresponse.." "..v2..","
else
completeresponse=completeresponse.." "..v1..v2.."."
end
end
response[1]=completeresponse
return response
else
return {[1]=""}
end
end
I don’t know if you find that useful but they are mostly pretty modular functions and are nice examples. I especially like the aware.judge.terrain function.
This humanoid description one is similar to my awareness functions in that it constructs a table of observations about the humanoid description and the first one is the complete observation.
My awareness library is coming along exceptionally! =] I really wanted the AI to know what they look like.
for sure, i’m thinking we’ll need help real soon, let me DM you
You asked for it, you waited for it… it’s finally here! 
…
NEW FEATURE - MULTIPLE AI NPCs
We’ve launched a new feature that will allow you to spin up multiple AI NPCs in one game, each with their own individual backstories and personalities! 
We’ve also fixed some bugs and cleaned up our UI quite a bit. We’re in final stages of testing some additional features such as putting AI NPCs behind a gamepass so you can start monetizing! 
Special thanks to @Magus_ArtStudios 
for helping out and coding some of these features
The feature is still in beta, so please bear with us! Happy creating! 
4 Likes
Yes I coded all these new features and tools for your to customize the AI with whatever you want! I use the most efficient methods to dynamically load assets of different types to the game via the AI’s various tools!
-AI Can Wear Clothes, hats, Tools from the Catalogue
-Can Change its Character via the catalouge or the Players Character
-Wikipedia Web_Tool that provides the AI with content filtered wikipedia articles and the AI provides a summary
-Memories the AI can remember you forever via ROBLOX datastores!
-Knows the time
-Efficiently load in playlists from the Catalogue.music or sound effects!
-AI knows player names
-Insert Emojis I used ChatGPT to create a accurate Emoji Insertion Model using various techniques and much data preparation including 1200+ Emojis!
Current Emojis Models (TinyEmoji and SmallEmoji)
-Game Pass or Group Pass Each AI can have different Abilities and so maybe you want to limit who can talk to them?
-Intelligent Emotes -I imported 169 animations from the Catalogue played each one and labeled carefully to create a expressive emote model using the ROBLOX created animations
-(In progress) The AI can create other NPCs currently but they do not have any combat AI inside.
I also added new developer Tools to the plugin allowing for Automatic UI scaling and easy way to add buttons of different options in addition to a module loader function, and a algorithm handling the variables of all the different features =]
4 Likes
seated issue is fixed in the latest update. you can say “follow me” your npc will jump up from seat.
text to speech is currently not allowed in roblox we will keep an eye on possible solution,
1 Like
It would be cool if several people could talk to the AI. Currently, only one person can talk to the AI
4 Likes