AI Wiki — Knowledge Base

We are in an insane run of open-weight drops. Every modality, open source is winning. This is what an open source AI summer ☀️ looks like: …

We are in an insane run of open-weight drops. Every modality, open source is winning. This is what an open source AI summer ☀️ looks like: 🧠 LLMs & Reasoning → DeepSeek-V4-Flash-0731 (my king 👑): 304B

model-releasesclem-delangue--x

12 Aug 2026

Search: “synthetic-data”

When Audit Quality Fails to Predict Downstream Utility: A Counterfactual Study of Synthetic-Data Selectors for Low-Resource African NLP

Source Identification Is Not Fitness Testing: Measuring the Limits of Synthetic-Data Attribution

Recursive LLM Degradation in Biomedical Question Answering: A Cross-Generation Study

Environment-free Synthetic Data Generation for API-Calling Agents

Unfolding Scientific Papers into Multi-Turn Generation Trajectories for Continued Pre-Training

SynCo: Data Synthesis Co-Training for Self-Evolving LLMs via Multi-Agent Reinforcement Learning

Broadly Applicable Approximate MCMC for Switching Stochastic Differential Equations Using Uniformization and Time-Conditioned Factorized Neural Likelihood Estimation

Concentration, Not Uncertainty: Why Targeted Synthetic Data Doesn't Help Camouflaged Object Detection

Hard, Yet Reducible: Controlled Forward Transfer for Synthetic Degradation Curation

From Behavior to Provenance: Attributing Tabular Foundation Models to Synthetic Pretraining Data

PaMIR: Open Benchmark of Public Credit-Default Datasets

Generation Provenance Before Behavior Attribution: Auditing Synthetic Speech Research Objects

Training-Aware Target Coverage for Synthetic Data Selection

Comparison of techniques for fine-tuning open-weight models for entity extraction from radiology reports

MaD-RL: Matching Distributions for Calibrating LLMs with Reinforcement Learning

Shared Autoregressive Context Can Distort Relationships in Synthetic Data

Synthetic data for ratemaking: imputation-based methods vs adversarial networks and autoencoders

Synthetic Thermal Image Generation for Real-Time Animal Detection Under Low-Visibility Conditions

AxonSynth: Domain-Randomized Synthetic Data for Zero-Shot 3D Axon Segmentation in Light-Sheet Microscopy

Learning Natural Conversational Behavior in Tandem Speech-to-Speech Models with Randomized Guidance

Transformer-based Monte Carlo Localization in Construction Meshes

Let Training Guide Selection: Online Synthetic Data Filtering via Real-Anchored Utility

Variational Quantum Transformer Architecture for Synthetic Language Generation

When Synthetic Data Hurts: On Catastrophic Forgetting in Skill Retrieval for LLM Agents

// Evaluating and Improving LLM Self-Modeling // Really interesting paper. Can a model answer questions about its own behavior? The question…

Building Pretraining Data for World Models: An Unreal Engine-Based Pipeline for Action-Conditioned Video Generation

Evaluating and Improving LLM Self-Modeling

FastVideo FastH3 V1: Open source 4-step Sparse Distilled H3 checkpoint/LORA

Mitigating LLM sycophancy with RL-based fine-tuning: Bayesian Truth Serum approach

Multiple View Neural Regression of a Facial Shape Model

The 10th AI City Challenge

We are in an insane run of open-weight drops. Every modality, open source is winning. This is what an open source AI summer ☀️ looks like: …

Entity-Faithful Repair of Synthetic Supervision for Zero-Shot Image Captioning

Real2Sim2Real for Vision-Language-Action Manipulation: An AMD ROCm-Based Pipeline

Automated Synthesis and Adversarial Validation of Executable Causal Research Pipelines

Language Generation with Replay: A Learning-Theoretic View of Model Collapse

TabQueryBench: A Query-Centric Benchmark for Synthetic Tabular Data

When Sample Selection Bias Precipitates Model Collapse

A Filtered Mixture-of-Generators for Fully Synthetic Survival Training

How Should World Models Be Evaluated for Embodied Decision-Making? A Decision-Making-Centric Position

Tractography-Driven Synthetic Data Generation for Fiber Bundle Segmentation in Tracer Histology

New research from Meta. Building synthetic training data has stayed a fixed pipeline that you hand-tune and then freeze. Autodata casts an A…

A specialized reasoning large language model for accelerating rare disease diagnosis: a randomized AI physician assistance trial

Enabling Robust Cloth Manipulation via Inference-Time Simulator-in-the-Loop Refinement

Object-Centric Dataset Resources for Constrained-Data Image Generation and Augmentation

When Web Agents Finish but Still Fail: Reproducible Triggers and Trace Diagnostics for Parallel Web Exploration

Bittensor Agent Arenas as a Trajectory Primitive: Distilling a Shopping Agent from ShoppingBench Subnet Traces

Train, Test, Re-evaluate: Schedule-Sensitive Evaluation of Generative Data for Hand Detection