Turns a single anime illustration into up to 23 semantically organized, inpainted layers with inferred drawing order, exported as a layered PSD. It provides a 2.5D animation starting point, but does not handle Live2D rigging or motion.
Turns research questions into reports through staged search, adversarial review, and citation checks, while saving sources in a persistent searchable vault. Runs with Claude Code or OpenAI Codex; full research can take hours.
Provides 55 million scene-level video clips (each with captions, language labels, and timestamps) extracted from an 80M-video, 10-million-hour raw pool to support multimodal pre-training across video, audio, and frames. Access is gated for academic/non-commercial research.
Tests whether language models can solve original, culturally grounded riddles—not just translate them. It spans 62 communities and combines model answers with native-speaker evaluation; use is restricted to evaluation.
Autonomously generates, iteratively debugs, and composes robot control programs into a reusable skill library for discovering and refining manipulation skills across simulation and real robots. Uses a code-as-policy continual learning loop—accumulating validated sensorimotor skills instead of model weights—to tackle long-horizon household and benchmark tasks.
Provides fixed-seed benchmark instances (prompts and agent-visible inputs) for ASI-Bench to run reproducible evaluations of LLM agents on scientific tasks. Includes four matched prompt levels (B1–B4) across 60 project-level tasks in 11 domains; excludes reference answers and private scorers; Apache-2.0 licensed.
Generated instance set (seed 31415) for ASI‑Bench: includes four matched prompt variants, agent-visible inputs, reference artifacts, and instance metadata for 60 project-scale scientific research tasks across 11 domains; intended for evaluating autonomous research agents. Licensed Apache‑2.0.
Treats agent self-improvement as natural selection over a population of harnesses (prompts, tools, skills, control flow), evolving a frozen-model agent by selecting harness edits that extend capability without regressing others. Uses a preserve-and-extend contract, lineage archive, and verifier-driven fitness (no gold solutions) to recombine complementary edits and transfer across benchmarks.
Provides ~39 TB of pre‑beamformed (channel capture) ultrasound RF data and metadata in zea/HDF5 format for reconstruction, flow, and inverse‑problem tasks. Released under CC‑BY‑4.0 and curated for training and evaluating ultrasound/RF foundation models.
Frames LLM routing as a sequential decision process and introduces LLMRouter plus the xRouteBench benchmark to develop, evaluate, and deploy learned routing policies across heterogeneous LLMs, optimizing response quality versus inference cost.
Evaluates whether LLM-driven storyteller agents preserve long-horizon logical consistency under adversarial player interventions. Introduces NCP-Bench (100 movie-derived narrative environments) with structured trajectory/commitments and automatic violation checks; finds strong LLMs often contradict themselves across multi-turn interactions.
Provides a year-scale multimodal benchmark and evaluation framework for on-device long-term memory in personal assistants, built from real mobile user trajectories. Tests memory construction, retrieval, updating, temporal reasoning, and implicit preference inference, and includes a knowledge-grounded synthesis pipeline to form coherent long-horizon trajectories.