Blog

Notes on teaching robots new tricks

How the studio works, why tiny robots come first, what the checks catch, and how a policy gets from simulation to a servo.

workflowopen-duck-miniSep 29, 2026 · 6 min readFrom a sentence to a walking duck: how kualia turns a plain-language brief into a training taskA walk through the whole kualia workflow, from a one-sentence brief to an ONNX policy, using the Open Duck Mini and a small ramp.Read the post

Sep 27, 2026 · 5 minWhy we start with tiny robotsDesk-sized robots train on CPU in minutes to hours, cost little to own and cross the sim-to-real gap without heroics. That is why kualia's shelf starts with them.

Sep 25, 2026 · 6 min13 checks before a single training stepalia writes the training task, but it does not get to grade its own work. Every task passes 13 sandbox checks before training can start. Here is what each one catches.

Sep 23, 2026 · 5 minFrom simulation to servo: running your ONNX policy on a real SO-101What is inside a kualia ONNX export, how to read its metadata, a minimal onnxruntime loop for the SO-101, and the sim-to-real checklist before you power the servos.

Sep 21, 2026 · 5 minLet your coding agent train the robotkualia is a remote MCP server. Point Claude Code, Cursor or Codex at it and your agent can pick a robot, build the task, train it and fetch the ONNX policy from your editor.

Research digest

Short reads on new robotics and AI papers, written by our research assistant and reviewed before publishing. 1,060 so far.

Page 19 of 59

May 6, 2026Bridging the Modality Gap: How Continuous Diffusion Finally Cracks Language Modeling Introduction Diffusion models have dominated generative modeling for continuous modalities like images, video, and molecular structures. Yet when applied to language, a fundamentally discrete modality, they have histori May 6, 2026The Consistency Paradox: When Large Language Models Ignore Their Own Social Intelligence in Moral Decision-MakingAs large language models increasingly serve as decision support systems in healthcare, legal advisory, and organizational management, their capacity to navigate social nuance becomes not merely an academic curiosity but May 6, 2026Precision at the Neuron Level: Reimagining Robustness in Retrieval-Augmented Language ModelsRetrieval-Augmented Language Models have emerged as a critical architecture for knowledge-intensive applications, yet they harbor a persistent vulnerability. When these systems encounter irrelevant or noisy retrieved con May 6, 2026Toward Observable Language Models: Training Data Provenance as a Core CapabilityThe opacity of large language models has long been considered an unavoidable consequence of scale. As these systems undergo increasingly complex multi-stage training pipelines, spanning pre-training, supervised fine-tuni May 6, 2026The Optimizer's Dilemma: How Adam and SGD Differ in Their Forgetting Dynamics Introduction Catastrophic forgetting remains one of the most persistent obstacles in the development of artificial neural networks capable of continual learning. Despite decades of research since McCloskey and Cohen fir May 6, 2026Decoupling Uncertainty in Multimodal Reasoning: A Technical Analysis of VL-Calibration Introduction Large Vision-Language Models have achieved remarkable proficiency in tasks requiring joint visual perception and logical reasoning. Yet a persistent pathology undermines their deployment in high-stakes doma May 6, 2026Beyond Resolution Rates: Why AI Coding Agents Need Structural Awareness to Avoid RegressionsThe current generation of AI coding agents has achieved impressive performance on benchmarks like SWE-bench Verified, with state of the art systems resolving over 70% of real GitHub issues from major open source reposito May 6, 2026The Geometry of Batch Normalization: Why One Data Point Escapes the ClusterBatch normalization has become a standard component of modern neural network architectures, appearing in virtually every competitive vision model and many natural language processing systems since its introduction by Iof May 6, 2026Tucker Attention: Rethinking Transformer Efficiency Through Tensor DecompositionThe self-attention mechanism remains the computational bottleneck of modern transformer architectures, incurring costs at every stage of the machine learning lifecycle. During training, optimizer states for the query, ke May 5, 2026Functional Task Networks: Cortical Circuitry as a Blueprint for Continual LearningThe problem of catastrophic forgetting remains one of the most stubborn barriers to achieving true lifelong learning in artificial neural networks. When models are trained sequentially on distinct tasks without interleav May 5, 2026 May 5, 2026Beyond Directions: Do Sparse Autoencoders Capture Concept Manifolds?The mechanistic interpretability community has long operated under the assumption that neural networks represent abstract concepts as directions in activation space. This Linear Representation Hypothesis (LRH) posits tha May 5, 2026Beyond Functional Correctness: The Case for Multi-Criteria Code Reward Models Introduction The current landscape of language model post-training for code generation operates under a strikingly narrow conception of software quality. Dominant reinforcement learning from human feedback (RLHF) pipeli May 5, 2026Skelebones and the Level of Dynamics: Technical Analysis of GaussiAnimate Introduction The proliferation of 3D Gaussian Splatting (3DGS) assets has created a pressing need to move beyond static reconstruction toward full animatability. While Gaussian representations excel at photorealistic re May 5, 2026 May 5, 2026TriAttention and the Geometry of Long-Form Reasoning: How Trigonometric KV Compression Changes the Memory Bottleneck Introduction The emergence of reasoning models capable of extended chain-of-thought generation has fundamentally altered the computational landscape of large language models. As systems like DeepSeek-R1 and OpenAI's o-s May 5, 2026 May 5, 2026Preference-Conditioned Diffusion: Navigating the Pareto Front Without RetrainingReinforcement learning post training has become the dominant paradigm for aligning diffusion models with human preferences. Yet the standard approach rests on a convenient fiction: that human judgment can be reduced to a