Measures how well LLMs and agent-driven workflows prepare supervised training data end-to-end by jointly benchmarking data construction and data-quality evaluation across six domains, using a downstream-grounded protocol and new metrics.
AI Dataset2026
Dataset of 5,000 reconstructed chain-of-thought samples produced by trace‑inversion from Claude‑opus‑4.7 summaries — packaged for SFT/DPO fine‑tuning. Key features: reconstructed CoT traces, multilingual prompts, gzip .jsonl format. Best used for reasoning distillation and model-level supervision; synthetic traces may need extra verification.
AI Dataset2026
Provides 9,000 reconstructed chain-of-thought (CoT) SFT examples produced by trace inversion from Claude Opus 4.6 outputs for fine-tuning reasoning-capable LLMs. Multilingual, packaged as .jsonl.gz and SFT/DPO-ready; verify numeric/code cases before training.
AI Dataset2026
RL training dataset for long-context language-model fine-tuning with ~23K samples and nine reward types, provided in Parquet with bilingual ground-truth and reward metadata for direct RL/bench evaluation.
AI Dataset2026
Supervised fine-tuning dataset of instruction-style examples in English and Chinese covering generation, QA, reasoning, math and code — targeted for SFT of 10–100B-parameter LLMs. Associated with arXiv:2602.09003; first published May 21, 2026.
AI Dataset2026
Parallel Chinese→Vietnamese dataset of webnovel (xianxia) text provided in JSON for NMT training and teacher-student distillation. In-domain, ~100K–1M examples with CC-BY-4.0 license — useful for fine-tuning or distillation experiments but limited by narrow genre and small download footprint.
AI Dataset2026
Provides raw newline-delimited JSON agent traces where assistant responses were generated by qwen/qwen3.7-max, captured with Teich; includes 47 JSONL files, an embedded tools schema snapshot, and conversion guidance for supervised fine‑tuning and distillation.
AI Dataset2026
Provides ~100 hours of expert-annotated, multi-channel Chinese conversational speech with per-segment timestamps, speaker IDs and paralinguistic labels for turn-taking, overlap/interruption detection and full‑duplex dialogue research. Licensed for academic/non-commercial use (CC BY‑NC 4.0).
Learns a text-conditioned flow (a conditional velocity field) in LLM residual activations to steer frozen models at inference by partially transporting and regenerating activations under target textual conditions — enabling unified control over persona, style, truthfulness, compositional constraints, and activation-space classification.
Introduces Draft-OPD, an on-policy distillation method for training lightweight draft models used in speculative decoding — it focuses learning on draft-induced errors via target-assisted rollouts and replay, improving acceptance length and enabling >5× lossless LLM inference acceleration.
AI Agent Papers2026
Analyzes when masking stale observations improves long-horizon search agents and why, identifying an asymmetric inverted-U relationship between masking benefit, retriever quality, and model capacity; explains a token-for-turn trade-off and releases evaluation scaffolds and trajectories.
AI Agent Papers2026
Automates distillation of heterogeneous traces from a target person or role into versioned, inspectable skill packages for LLM agents — producing separate capability and bounded-behavior tracks that support natural-language corrections, rollback, and cross-host installation. Ships with an open system and a skills gallery.