datasets
Training and evaluation data, with the modality, task and licence stated up front. Listed live from the Hugging Face Hub.
Recursive-Task-Synthesis
Recursive Task Synthesis
This dataset contains 37,484 validated command-line task instances produced
through recursive task synthesis. Public identifiers are opaque and stable.
metadata/tasks.parquet: one searchable row per task instance.
metadata/shard_manifest.jsonl: TAR sizes and SHA256 checksums.
data/tasks-*.tar: sanitized runnable task packages.
The searchable task rows include:
instruction: contents of instruction.md.
task_toml: contents of task.toml.
solution:… See the full description on the dataset page: https://huggingface.co/datasets/Zhongzhi1228/Recursive-Task-Synthesis.Recursive-Task-Synthesis-Trajectories
Recursive Task Synthesis Trajectories
This dataset contains 327,189 completed agent trajectories collected on
recursively synthesized command-line tasks. Public identifiers are opaque and
stable.
The trajectory JSON retains messages, actions, observations, and token counts.
Token-level log-probability arrays and duplicated debug/session captures are
excluded from the public packages.
metadata/trajectories.parquet: searchable trajectory metadata.
metadata/shard_manifest.jsonl:… See the full description on the dataset page: https://huggingface.co/datasets/Zhongzhi1228/Recursive-Task-Synthesis-Trajectories.recursive-task-synthesis-glm-5.3-rollouts
GLM 5.3 agentic rollouts on Recursive-Task-Synthesis
This dataset catalogs the full collection made from the pinned
Recursive-Task-Synthesis dataset revision
be44f96808d5a9b599d5cb024341ff00091adeb7. The repository includes approximately 260.5 GiB of trajectory payload tar shards.
Contents at a glance
Item
Count
Source tasks considered
37,284
Source candidates inspected
19,368
Converted tasks after source filters
18,600
Tasks passing gold… See the full description on the dataset page: https://huggingface.co/datasets/open-athena/recursive-task-synthesis-glm-5.3-rollouts.Recursive-Task-Synthesis-Quality-1K
Recursive Task Synthesis Quality 1K
This dataset contains 1,000 quality-selected, validated command-line task
instances. It is a curated subset of the
Recursive Task Synthesis dataset.
Public task and group identifiers are opaque and stable across both datasets.
Selection
The subset was selected from 37,484 validated tasks using structural and safety
checks, two-pass semantic review, strict gates for instruction clarity,
instruction-verifier alignment, verifier… See the full description on the dataset page: https://huggingface.co/datasets/Zhongzhi1228/Recursive-Task-Synthesis-Quality-1K.Recursive-Task-Synthesis
Recursive Task Synthesis
Only tasks with completed PUBLIC platform container and VM artifacts at the
2026-09-17T02:53:50.654736+00:00 registry audit are included. The excluded IDs, reasons
and last failed build IDs are recorded in exclusions.json. Exclusions affect
both metadata rows and complete TAR task packages. This is a hosting filter,
not gold validation. Recheck both artifact types and remove recovered IDs from
the preparation sidecar to restore them from the pinned… See the full description on the dataset page: https://huggingface.co/datasets/PrimeIntellect/Recursive-Task-Synthesis.Recursive-Task-Synthesis-Copy
Recursive Task Synthesis
This dataset contains 37,484 validated command-line task instances produced
through recursive task synthesis. Public identifiers are opaque and stable.
metadata/tasks.parquet: one searchable row per task instance.
metadata/shard_manifest.jsonl: TAR sizes and SHA256 checksums.
data/tasks-*.tar: sanitized runnable task packages.
The searchable task rows include:
instruction: contents of instruction.md.
task_toml: contents of task.toml.
solution:… See the full description on the dataset page: https://huggingface.co/datasets/AmanPriyanshu/Recursive-Task-Synthesis-Copy.qwen3-4b-recursive-sft-v3.1-eval-rollouts
Recursive V3.1 SFT Evaluation Rollouts
Saved evaluation generations from
tawer12/qwen3-4b-recursive-sft-v3.1.
This dataset contains SFT-model evaluations only, not RL training rollouts,
flattened-model generations, or the SFT training corpus.
The release preserves all original per-rollout fields and text, including
invalid trees and wrong answers. Added fields identify the benchmark, attempt,
model, and source file/line. No generations or answer labels were repaired.… See the full description on the dataset page: https://huggingface.co/datasets/tawer12/qwen3-4b-recursive-sft-v3.1-eval-rollouts.samantha-r01-recursive-reasoning-corpus
Samantha R01 Recursive Reasoning Corpus
Answer-only corpus for the first isolated Samantha silent-tick / recursive-latent-reasoning validation. It is normalized for pre_train_recursive_reasoning.py and intentionally contains no visible chain-of-thought or source rationale fields.
The repository is private because it combines sources with mixed or unspecified redistribution terms. Access does not supersede any upstream license.
Split policy
train: 50,000… See the full description on the dataset page: https://huggingface.co/datasets/BRlkl/samantha-r01-recursive-reasoning-corpus.GutenQA_Recursive
📚 GutenQA-Recursive
GutenQA-Recursive consists on the same 100 Public Domain Narrative Books used in GutenQA (the proposed benchmark to the paper LumberChunker: Long-Form Narrative Document Segmentation, and serves as one of the baseline chunking approaches utilized on the LumberChunker paper.
In this version, passages are segmented with Langchain's Recursive Character Splitting function.
The dataset is organized into the following columns:
Book Name: The title of the book from… See the full description on the dataset page: https://huggingface.co/datasets/LumberChunker/GutenQA_Recursive.maze-navigation-6-types-kruskal-recursive