datasets
Training and evaluation data, with the modality, task and licence stated up front. Listed live from the Hugging Face Hub.
world-model-pi-sft-formatted
world-model-pi-sft-formatted
Role-flipped formatted SFT data for training a world model (environment simulator) over the
TeichAI pi coding-agent harness. Source traces: TeichAI/DeepSeek-v4-Pro-Agent (4006 sessions).
This dataset is stage 1 of a two-stage pipeline: stage 2 (world-model-pi-sft-with-cot)
injects synthetic <think> rationales into every assistant target via a reasoning-model
teacher; this stage 1 dataset has only the placeholder <COT_PLACEHOLDER> token that stage 2… See the full description on the dataset page: https://huggingface.co/datasets/kfallah/world-model-pi-sft-formatted.babyai-world-model-sft
BabyAI World Model SFT Dataset
Training data for supervised fine-tuning (SFT) of a world model for the BabyAI grid-world environment. The world model learns to predict the next observation and available actions given the current state and the agent's action.
Dataset Description
Each example is a single environment transition in chat format (system/user/assistant):
System: Detailed simulator prompt with environment rules and task completion logic
User: Goal, current… See the full description on the dataset page: https://huggingface.co/datasets/GGOSinon/babyai-world-model-sft.