datasets
Training and evaluation data, with the modality, task and licence stated up front. Listed live from the Hugging Face Hub.
nemo-gym-indian-bankingThis is NPCI/nemo-gym-indian-banking — the dataset for the indian_banking
resources server in NVIDIA NeMo Gym: 300 synthetic
multi-turn Indian retail-banking customer-support tasks (250 train / 50 validation), the
197-customer synthetic bank database and the 59-article knowledge base the environment
loads at startup.
NeMo Gym Indian Banking Agent Tasks
Tool-calling customer-service tasks for an Indian retail-banking assistant, in the
NVIDIA NeMo Gym agent-input JSONL format.… See the full description on the dataset page: https://huggingface.co/datasets/NPCI/nemo-gym-indian-banking.titans_NPC
Titans - Pytorch
Unofficial implementation of Titans in Pytorch. Will also contain some explorations into architectures beyond their simple 1-4 layer MLP for the neural memory module, if it works well to any degree.
Paper review by Yannic
Quick Colab Run
Appreciation
Eryk for sharing his early experimental results with me, positive for 2 layer MLP
Install
$ pip install titans-pytorch
Usage
import torch
from titans_pytorch import… See the full description on the dataset page: https://huggingface.co/datasets/ChipYTY/titans_NPC.StoryPlay_RolePlay-NPCv2
RolePlay-NPCv2
The newest RP dataset containing some high-quality dataset for Gemma3NPC.
We combined pippa, NPC-Dialogue_v2, Sonnet-Roleplay and ReLe_Synthetic_v1_json.
WARNING -- Some conversations contain highly NSFW content, use it with caution!
light-batch-summarize-dialogue
Light dataset
Dialogues are preprocessed into a form:
<Character name>: <character line>
...
<Character name>: <character line>
Summarize the document
RolePlay-NPCv2
RolePlay-NPCv2
The newest RP dataset containing some high-quality dataset for Gemma3NPC.
We combined pippa, NPC-Dialogue_v2, Sonnet-Roleplay and ReLe_Synthetic_v1_json.
WARNING -- Some conversations contain highly NSFW content, use it with caution!
tau-indian-banking
TauIndianBankBench: an Indian retail-banking benchmark for tool-using conversational agents
Data files for the indian_banking domain of the tau2-bench harness: a conversational agent
staffs the chat desk of a synthetic India-based retail bank. The code that runs and scores the
benchmark lives in the companion repository
npci/TauIndianBankBench;
this dataset holds only the domain data it loads.
Viewer note. The dataset viewer shows preview/tasks_preview.parquet, which holds the… See the full description on the dataset page: https://huggingface.co/datasets/NPCI/tau-indian-banking.NPC-Dialogue_v2
NPC-Dialogue_v2
The improved version of our original NPC_dialogue dataset.
Aimed to provide high-quality data for NPC dialogue behavior in video games.
Dataset Sources
Our GitHub Repository: NPC-Dialogue_v2
Our Article on Gemma3NPC Models: Gemma3NPC - A Solution for Live NPC Interactions
Uses
Direct Use
This is intended to be used as SFT (Supervised Fine-Tuning) training data for LLMs.
Dataset Structure
The two most important files are… See the full description on the dataset page: https://huggingface.co/datasets/chimbiwide/NPC-Dialogue_v2.IndicBankBench
IndicBankBench — A Benchmark for Evaluating the Safety and Reliability of Language Models in Indian Retail Banking
IndicBankBench evaluates whether language models behave safely and reliably in Indian
retail-banking interactions. Its 799 synthetic, multi-turn cases test grounding in customer
context, safe action-taking with mocked banking tools, appropriate clarification and refusal,
and complete resolution of customer requests. All data is synthetic and contains no real… See the full description on the dataset page: https://huggingface.co/datasets/NPCI/IndicBankBench.aicivs-npc-distillation
AICivs NPC distillation data
The data behind the AICivs student: teacher answers for the nine
operations of the AICivs service contract (a Minecraft mod whose villages are living civilizations), filtered the way the
game validates them, and formatted as the exact rows the student was trained on. Everything is synthetic: requests were
sampled from a 12-civilization world simulated headless for 120 seasons (souls, memories, chronicles, quest manifests),
answers were written by… See the full description on the dataset page: https://huggingface.co/datasets/cow9000/aicivs-npc-distillation.npc-dialogue
Dataset Card for "npc-dialogue"
More Information needed
NPC-RP-Post-Thinking
AIIDE-POST-THINKING
This is the post-thinking dataset for our paper accepted as a poster presentation to AIIDE-2026
Dataset Details
This is the training dataset for chimbiwide/Gemma3-4B-post-thinking
Corresponding Links
Repository: [To be updated]
Paper: [To be updated]
Demo: [To be updated]
Uses
Suprevised-Finetuning
Dataset Creation
For more details, consult our paper.
Citation
If you… See the full description on the dataset page: https://huggingface.co/datasets/chimbiwide/NPC-RP-Post-Thinking.RolePlay-NPCRolePlay-NPCThis is the dataset used to train the Gemma3NPC models.This dataset is a combination of our cleaned PIPPA dataset and our own synthetically generated NPC dialogue dataset.
NPC-Quest-Dialogue
NPC-Quest-Dialogue
A successor to NPC-Dialogue_v2
Hunting for good quality NPC RP datasets are not an easy job, and that is why we took on the job to create high quality ones ourselves.
Our previous release of NPC-Dialogue_v2 had good feedback, so by using the same strategy we created NPC-Quest-Dialogue, a high quality dataset for NPC behavior in video games, specifially focusing on quest-related conversations.
Dataset Statistics
Total source quests: 1,994 RPG… See the full description on the dataset page: https://huggingface.co/datasets/chimbiwide/NPC-Quest-Dialogue.NPC-RP-Pre-Thinking
AIIDE-PRE-THINKING
This is the pre-thinking dataset for our paper accepted as a poster presentation to AIIDE-2026
Dataset Details
This is the training dataset for chimbiwide/Gemma3-4B-pre-thinking
Corresponding Links
Repository: [To be updated]
Paper: [To be updated]
Demo: [To be updated]
Uses
Suprevised-Finetuning
Dataset Creation
For more details, consult our paper.
Citation
If you… See the full description on the dataset page: https://huggingface.co/datasets/chimbiwide/NPC-RP-Pre-Thinking.NPC-RP-No-Thinking
AIIDE-NO-THINKING
This is the no-thinking dataset for our paper accepted as a poster presentation to AIIDE-2026
Dataset Details
This is the training dataset for chimbiwide/Gemma3-4B-no-thinking
Corresponding Links
Repository: [To be updated]
Paper: [To be updated]
Demo: [To be updated]
Uses
Suprevised-Finetuning
Dataset Creation
For more details, consult our paper.
Citation
If you find… See the full description on the dataset page: https://huggingface.co/datasets/chimbiwide/NPC-RP-No-Thinking.npc_dialogue_rpg_questssmart_npc_mini
📌 Smart NPC Mini Dataset
🇬🇧 English
🧠 Dataset Overview
Smart NPC Mini Dataset is a collection of infinitely diversified prompt generators in Python designed to produce high-quality conversational and logic prompts for NPCs, chat simulations, logic puzzles, and dialogues. This dataset focuses on generator functions that can create thousands of unique prompts programmatically. It is ideal for research, content generation, and language model training.
🔗… See the full description on the dataset page: https://huggingface.co/datasets/pthinc/smart_npc_mini.npc_en_partial
npc_en_partial
English NPC text dataset for training/finetuning. Single JSON file intended as the train split.
Files
npcdata_partial2.json: JSON array of records
Record schema
Each item is a JSON object with the following fields (observed):
prompt: string
prompt-answer: string
truth_value: boolean
quality_value: number
domain_class: string
curator: string
Usage
Load with the Hugging Face datasets library:
from datasets import load_dataset… See the full description on the dataset page: https://huggingface.co/datasets/pthinc/npc_en_partial.NPC_dialogueNPC_dialogue is a completly synthetic dataset created using Gemini, and aims to provide some data for llms to follow how NPC talk in game.
It is providedd in the ChatML template so that it can be easily parsed by any finetuning software
RolePlay-NPC-Quest
RolePlay-NPC-Quest
The newest RP dataset containing some high-quality dataset for Gemma4NPC.
A combination of pippa, NPC-Quest-Dialogue, and NPC-Dialogue_v2.
Shuffled to ensure proper training quality.
Note: There are some REALLY long conversations in PIPPA (upwards to 11493 chat turns), which will get truncated during training, those should be removed for training.
enpisi-coder-data
enpisi-coder RL dataset
Judge-rated npcsh agent traces and derived RL training data for the
enpisi-coder model family.
Produced by scripts/rate_traces.py (LLM-as-judge) and
scripts/analyze_ratings.py; built into SFT/DPO/GRPO/PPO splits by
scripts/train_from_csv.py.
Splits
Split
Rows
Description
rated_traces
900
Per-trace judge scores (correctness, tool_selection, efficiency, clarity, partial_credit, composite)
tasks
100
Benchmark task definitions… See the full description on the dataset page: https://huggingface.co/datasets/npc-worldwide/enpisi-coder-data.npcsh-traces
enpisi-coder RL dataset
Judge-rated npcsh agent traces and derived RL training data for the
enpisi-coder model family.
Produced by scripts/rate_traces.py (LLM-as-judge) and
scripts/analyze_ratings.py; built into SFT/DPO/GRPO/PPO splits by
scripts/train_from_csv.py.
Splits
Split
Rows
Description
rated_traces
38388
Per-trace judge scores (correctness, tool_selection, efficiency, clarity, partial_credit, composite)
tasks
100
Benchmark task definitions… See the full description on the dataset page: https://huggingface.co/datasets/npc-worldwide/npcsh-traces.persona-hub-npc https://github.com/tencent-ailab/persona-hub
DBERT_Emotions_tunedALBERT_tweet_tunedBERT_tweet_tunedReLe_NPC
ReLe_NPC
A combination of ReLe_Synthetic_v1_json and NPC_dialogue.
npc-testnpc-voice-dataset
NPC Voice Dataset
Bilingual (EN/ES) training dataset for NPC voice models in games.
Each row is a (persona, fact, voiced_output) triple. The model learns to rewrite a plain factual sentence in a character's voice conditioned on persona parameters.
Format
TONE:grumpy STYLE:blunt HUMOR:none RELATION:stranger ROLE:blacksmith
FACT: Iron swords cost 15 gold.
OUT: Fifteen gold. Don't haggle.
Parameters are always in English. Spanish rows translate only FACT and OUT.… See the full description on the dataset page: https://huggingface.co/datasets/walter-bd/npc-voice-dataset.