task_types listlengths 1 4 | difficulty large_stringclasses 4
values | conversations listlengths 4 10 | processed large_stringlengths 1.38k 17.1k |
|---|---|---|---|
[
"Bug Fix",
"Q/A"
] | beginner | [
{
"role": "user",
"content": "My Studio recipe-checking utility has a verbose option. This makes it verbose even when I pass false. I only want an omitted option to default to true.\n```luau\nlocal function checkRecipes(options)\n local verbose = options.verbose or true\n if verbose then\n prin... | <|open|>message role="system" type="thinking-effort"<|sep|>`thinking_effort` guides on how much to think in your thinking channel (not including the response channel), supported values include `low`, `medium`, `high`, and `max`.
Now the system is invoked with `thinking_effort=max`.<|close|>message<|sep|><|end_of_msg|><... |
[
"Debugging",
"Bug Fix"
] | beginner | [
{
"role": "user",
"content": "I'm inspecting numbered crafting slots in a Roblox ModuleScript. There are three occupied slots, but my count and loop don't agree with that.\n```luau\nlocal slots = { [1] = \"Wood\", [4] = \"Stone\", [9] = \"Fiber\" }\nprint(#slots)\nfor slot, item in ipairs(slots) do\n pri... | <|open|>message role="system" type="thinking-effort"<|sep|>`thinking_effort` guides on how much to think in your thinking channel (not including the response channel), supported values include `low`, `medium`, `high`, and `max`.
Now the system is invoked with `thinking_effort=max`.<|close|>message<|sep|><|end_of_msg|><... |
[
"Bug Fix",
"Optimization"
] | intermediate | [
{
"role": "user",
"content": "This function in my recipe-job ModuleScript sometimes leaves expired jobs behind, especially two adjacent ones. `jobs` is a dense array, and `now` is passed in so I can test it.\n```luau\nlocal function prune(jobs, now)\n for index, job in ipairs(jobs) do\n if job.exp... | <|open|>message role="system" type="thinking-effort"<|sep|>`thinking_effort` guides on how much to think in your thinking channel (not including the response channel), supported values include `low`, `medium`, `high`, and `max`.
Now the system is invoked with `thinking_effort=max`.<|close|>message<|sep|><|end_of_msg|><... |
[
"Debugging",
"Bug Fix",
"Refactor"
] | intermediate | [
{
"role": "user",
"content": "My Roblox quest simulation tests leak progress between runs. The QuestProgress ModuleScript has this factory:\n```luau\nlocal defaults = {\n coins = 0,\n quests = { Intro = { step = 0, done = false } },\n}\nlocal function newProgress()\n return table.clone(defaults)\ne... | <|open|>message role="system" type="thinking-effort"<|sep|>`thinking_effort` guides on how much to think in your thinking channel (not including the response channel), supported values include `low`, `medium`, `high`, and `max`.
Now the system is invoked with `thinking_effort=max`.<|close|>message<|sep|><|end_of_msg|><... |
[
"Review",
"Bug Fix",
"Q/A"
] | intermediate | [
{
"role": "user",
"content": "Please review this tiny Catalog ModuleScript for runtime immutability. This is the entire schema; power is a number.\n```luau\nlocal catalog = table.freeze({\n Hammer = { power = 12 },\n})\nreturn catalog\n```\nMy recipe editor shouldn't be able to change any shared catalog ... | <|open|>message role="system" type="thinking-effort"<|sep|>`thinking_effort` guides on how much to think in your thinking channel (not including the response channel), supported values include `low`, `medium`, `high`, and `max`.
Now the system is invoked with `thinking_effort=max`.<|close|>message<|sep|><|end_of_msg|><... |
[
"Debugging",
"Bug Fix",
"Q/A"
] | beginner | [
{
"role": "user",
"content": "I'm counting validated assets with a Counter ModuleScript. This is in ServerScriptService next to my test Script.\n```luau\nlocal Counter = { value = 0 }\nfunction Counter:add(amount)\n self.value += amount\nend\nreturn Counter\n```\nThe test calls it like this and gets an a... | <|open|>message role="system" type="thinking-effort"<|sep|>`thinking_effort` guides on how much to think in your thinking channel (not including the response channel), supported values include `low`, `medium`, `high`, and `max`.
Now the system is invoked with `thinking_effort=max`.<|close|>message<|sep|><|end_of_msg|><... |
[
"Bug Fix",
"Q/A"
] | advanced | [
{
"role": "user",
"content": "My Studio command recorder has optional positional arguments, including trailing nils. This loses the argument count:\n```luau\nlocal function capture(...)\n return { ... }\nend\nlocal saved = capture(\"Rename\", nil, 3, nil)\nprint(#saved)\n```\nHow do I preserve all four p... | <|open|>message role="system" type="thinking-effort"<|sep|>`thinking_effort` guides on how much to think in your thinking channel (not including the response channel), supported values include `low`, `medium`, `high`, and `max`.
Now the system is invoked with `thinking_effort=max`.<|close|>message<|sep|><|end_of_msg|><... |
[
"Bug Fix",
"Q/A"
] | beginner | [
{
"role": "user",
"content": "In a strict Roblox building-tool module, this helper complains that Transparency isn't on Instance, and that candidate might be nil:\n```luau\n--!strict\nlocal function paintMarker(candidate: Instance?)\n candidate.Transparency = 0.5\nend\n```\nA marker is supposed to be a P... | <|open|>message role="system" type="thinking-effort"<|sep|>`thinking_effort` guides on how much to think in your thinking channel (not including the response channel), supported values include `low`, `medium`, `high`, and `max`.
Now the system is invoked with `thinking_effort=max`.<|close|>message<|sep|><|end_of_msg|><... |
[
"Generation",
"Q/A"
] | intermediate | [
{
"role": "user",
"content": "Can you write a strict StackCount ModuleScript for my Roblox inventory-data validator? parse(text) should accept counts 1 through 999, integers only, and return either a number or an explanation. Normal Luau tonumber syntax is fine. I'd prefer a tagged result instead of throwin... | <|open|>message role="system" type="thinking-effort"<|sep|>`thinking_effort` guides on how much to think in your thinking channel (not including the response channel), supported values include `low`, `medium`, `high`, and `max`.
Now the system is invoked with `thinking_effort=max`.<|close|>message<|sep|><|end_of_msg|><... |
[
"Debugging",
"Bug Fix",
"Refactor"
] | intermediate | [
{
"role": "user",
"content": "This pure helper is used by my Studio asset-name importer. Invalid size text is supposed to produce nil and a message, but the pcall says success and then I get an arithmetic error.\n```luau\nlocal function parseSize(text)\n local value = tonumber(text)\n if value == nil ... | <|open|>message role="system" type="thinking-effort"<|sep|>`thinking_effort` guides on how much to think in your thinking channel (not including the response channel), supported values include `low`, `medium`, `high`, and `max`.
Now the system is invoked with `thinking_effort=max`.<|close|>message<|sep|><|end_of_msg|><... |
[
"Bug Fix",
"Refactor"
] | intermediate | [
{
"role": "user",
"content": "My Roblox build-manifest exporter sorts entries by priority, highest first. Priorities are finite numbers and IDs are unique strings. Equal-priority entries move around when the input order changes, so exports get noisy diffs.\n```luau\ntable.sort(entries, function(a, b)\n r... | <|open|>message role="system" type="thinking-effort"<|sep|>`thinking_effort` guides on how much to think in your thinking channel (not including the response channel), supported values include `low`, `medium`, `high`, and `max`.
Now the system is invoked with `thinking_effort=max`.<|close|>message<|sep|><|end_of_msg|><... |
[
"Generation",
"Q/A"
] | intermediate | [
{
"role": "user",
"content": "I need an AssetChoice ModuleScript in ReplicatedStorage for selecting an asset ID from a dense array. Empty pools should return nil, not error. Please accept a Roblox Random object so tests can supply their own seed, and don't modify the pool. A generic typed function would be ... | <|open|>message role="system" type="thinking-effort"<|sep|>`thinking_effort` guides on how much to think in your thinking channel (not including the response channel), supported values include `low`, `medium`, `high`, and `max`.
Now the system is invoked with `thinking_effort=max`.<|close|>message<|sep|><|end_of_msg|><... |
[
"Generation",
"Q/A"
] | advanced | [
{
"role": "user",
"content": "Write a pure MapStyleWeights ModuleScript in ServerScriptService. Input is a dense array of {id: string, weight: number}. Nonnegative integer weights only, and cap total tickets at 1,000,000 as our own data rule. Zero-weight styles must never win. Empty/all-zero pools return ni... | <|open|>message role="system" type="thinking-effort"<|sep|>`thinking_effort` guides on how much to think in your thinking channel (not including the response channel), supported values include `low`, `medium`, `high`, and `max`.
Now the system is invoked with `thinking_effort=max`.<|close|>message<|sep|><|end_of_msg|><... |
[
"Debugging",
"Refactor"
] | intermediate | [
{
"role": "user",
"content": "My Studio source-analysis module uses session tables as weak cache keys. Dropping the session variable doesn't seem to release its entry while work is pending. This is the relevant code; pending is a normal module-local table, not task.spawn work.\n```luau\nlocal cache = setmet... | <|open|>message role="system" type="thinking-effort"<|sep|>`thinking_effort` guides on how much to think in your thinking channel (not including the response channel), supported values include `low`, `medium`, `high`, and `max`.
Now the system is invoked with `thinking_effort=max`.<|close|>message<|sep|><|end_of_msg|><... |
[
"Debugging",
"Refactor",
"Bug Fix"
] | advanced | [
{
"role": "user",
"content": "My item-description utility stops during require. These are sibling ModuleScripts in ServerScriptService.ItemTools. Is returning the table earlier or adding a wait the right fix?\n```luau\n-- ItemBook\nlocal FormatItem = require(script.Parent.FormatItem)\nlocal ItemBook = {}\nI... | <|open|>message role="system" type="thinking-effort"<|sep|>`thinking_effort` guides on how much to think in your thinking channel (not including the response channel), supported values include `low`, `medium`, `high`, and `max`.
Now the system is invoked with `thinking_effort=max`.<|close|>message<|sep|><|end_of_msg|><... |
[
"Q/A",
"Generation"
] | beginner | [
{
"role": "user",
"content": "ReplicatedStorage.BuildCounter is a ModuleScript containing `return { value = 0 }`. An ordinary server Script requires it and sets value to 10. My LocalScript requires the same ModuleScript but still reads 0. Neither uses Actors. I thought require caches one table, so why isn't... | <|open|>message role="system" type="thinking-effort"<|sep|>`thinking_effort` guides on how much to think in your thinking channel (not including the response channel), supported values include `low`, `medium`, `high`, and `max`.
Now the system is invoked with `thinking_effort=max`.<|close|>message<|sep|><|end_of_msg|><... |
[
"Bug Fix",
"Generation"
] | beginner | [
{
"role": "user",
"content": "I'm making plain track records for a Roblox timeline-authoring utility, not creating Animation instances yet. Why does renaming track 1 rename every track?\n```luau\nlocal tracks = table.create(3, { name = \"\", keyframes = {} })\ntracks[1].name = \"Opening\"\nassert(tracks[2].... | <|open|>message role="system" type="thinking-effort"<|sep|>`thinking_effort` guides on how much to think in your thinking channel (not including the response channel), supported values include `low`, `medium`, `high`, and `max`.
Now the system is invoked with `thinking_effort=max`.<|close|>message<|sep|><|end_of_msg|><... |
[
"Bug Fix",
"Q/A"
] | beginner | [
{
"role": "user",
"content": "My Roblox building-tool module has a pure coordinate snap helper. Inputs are finite coordinates from -10000 to 10000; supported steps are 0.25, 1, and 2. Negative halfway coordinates snap toward zero, which looks asymmetric.\n```luau\nlocal function snapCoordinate(value: number... | <|open|>message role="system" type="thinking-effort"<|sep|>`thinking_effort` guides on how much to think in your thinking channel (not including the response channel), supported values include `low`, `medium`, `high`, and `max`.
Now the system is invoked with `thinking_effort=max`.<|close|>message<|sep|><|end_of_msg|><... |
[
"Bug Fix",
"Generation"
] | intermediate | [
{
"role": "user",
"content": "My Studio asset-renaming plugin uses this local helper on names. I only need the returned string for now. Searching for Ore[Old].v1 doesn't match that literal text, and replacement names can contain percent signs. Empty search strings should be rejected.\n```luau\nlocal functio... | <|open|>message role="system" type="thinking-effort"<|sep|>`thinking_effort` guides on how much to think in your thinking channel (not including the response channel), supported values include `low`, `medium`, `high`, and `max`.
Now the system is invoked with `thinking_effort=max`.<|close|>message<|sep|><|end_of_msg|><... |
[
"Review",
"Q/A"
] | beginner | [
{
"role": "user",
"content": "Please review this lookup in my Roblox achievement-import utility. Missing IDs are intentionally nil and IDs are case-sensitive strings. The table is local, plain, and initialized before calls.\n```luau\n--!strict\ntype Achievement = { label: string, unlocked: boolean }\nlocal ... | <|open|>message role="system" type="thinking-effort"<|sep|>`thinking_effort` guides on how much to think in your thinking channel (not including the response channel), supported values include `low`, `medium`, `high`, and `max`.
Now the system is invoked with `thinking_effort=max`.<|close|>message<|sep|><|end_of_msg|><... |
[
"Debugging",
"Bug Fix"
] | beginner | [
{
"role": "user",
"content": "My Roblox banner selector does nothing. ChooseBanner is a RemoteEvent in ReplicatedStorage. The LocalScript in StarterPlayerScripts sends this:\n```luau\nChooseBanner:FireServer(game.Players.LocalPlayer, \"Blue\")\n```\nThe Script in ServerScriptService has this handler. The pr... | <|open|>message role="system" type="thinking-effort"<|sep|>`thinking_effort` guides on how much to think in your thinking channel (not including the response channel), supported values include `low`, `medium`, `high`, and `max`.
Now the system is invoked with `thinking_effort=max`.<|close|>message<|sep|><|end_of_msg|><... |
[
"Review",
"Bug Fix",
"Q/A"
] | intermediate | [
{
"role": "user",
"content": "Review the quantity check in my session-only ticket exchange. In a ServerScriptService Script, ExchangeTickets.OnServerEvent supplies player and quantity. A non-yielding admission check already limits requests to four per second per player. profiles[player] is server-owned, exi... | <|open|>message role="system" type="thinking-effort"<|sep|>`thinking_effort` guides on how much to think in your thinking channel (not including the response channel), supported values include `low`, `medium`, `high`, and `max`.
Now the system is invoked with `thinking_effort=max`.<|close|>message<|sep|><|end_of_msg|><... |
[
"Review",
"Bug Fix"
] | intermediate | [
{
"role": "user",
"content": "Can you review this small cosmetic unlock endpoint? BuyStyle is a RemoteEvent in ReplicatedStorage. This is server code; allowPurchase is an existing non-yielding per-player limiter, and sessions is a server table cleaned on PlayerRemoving. Each session has coins and an unlocke... | <|open|>message role="system" type="thinking-effort"<|sep|>`thinking_effort` guides on how much to think in your thinking channel (not including the response channel), supported values include `low`, `medium`, `high`, and `max`.
Now the system is invoked with `thinking_effort=max`.<|close|>message<|sep|><|end_of_msg|><... |
[
"Bug Fix",
"Q/A"
] | intermediate | [
{
"role": "user",
"content": "I have a logical loadout, not Tool instances. ReplicatedStorage.ItemDefinitions contains direct Configuration children named Lantern and Compass, with unique names. Profiles are server-only tables: profiles[player] = {owned = {Lantern = true}, equipped = nil}. SetLoadoutItem is... | <|open|>message role="system" type="thinking-effort"<|sep|>`thinking_effort` guides on how much to think in your thinking channel (not including the response channel), supported values include `low`, `medium`, `high`, and `max`.
Now the system is invoked with `thinking_effort=max`.<|close|>message<|sep|><|end_of_msg|><... |
[
"Bug Fix",
"Generation"
] | intermediate | [
{
"role": "user",
"content": "My lobby chime has a five-second LocalScript cooldown, but our tester can trigger it repeatedly by calling the remote directly. RequestChime and ChimePlayed are RemoteEvents in ReplicatedStorage. Currently the server just fires ChimePlayed to everyone with the sender's UserId. ... | <|open|>message role="system" type="thinking-effort"<|sep|>`thinking_effort` guides on how much to think in your thinking channel (not including the response channel), supported values include `low`, `medium`, `high`, and `max`.
Now the system is invoked with `thinking_effort=max`.<|close|>message<|sep|><|end_of_msg|><... |
End of preview. Expand in Data Studio
Luau-Coder 1.0 Preview SFT
This dataset is exceptionally high-quality supervised fine-tuning conversations for a highly capable coding model in Roblox Luau domain. It prioritize technical correctness, useful engineering judgment, realistic interaction, and efficient explanations over output volume.
This dataset includes:
- Multi-turns (2-5 turns)
- Dynamic CoT (length)
- Dynamic Interleaved Reasoning
- Q/A
- Review
- Debugging
- Bug Fix
- Refactor
- Optimization
- Code Generation
- Ideas
License
Apache 2.0
This dataset is NOT completed, and temporary not human verified yet. This dataset is NOT designed and targeted for Large Model scale due to it averages CoT length.
Generated by OpenAI GPT-6 Astra (high) for the first 506 rows (cost $90 USD). Using cheaper model to continue generates this since it have enough of samples as good references
- Downloads last month
- 43