ZK
Editor-in-Chief · CPO of AlpinaGPT

Zhemal Khamidun

Head of AI at Alpina Digital, CPO of AlpinaGPT (~30-person engineering & AI team). Founder of Vibe Academy & Hamidun Consulting, author of the AI channel «Готовим ИИшницу» (@jemal_hamidun).

10,682Published articles
50+Verified primary sources
6Language editions (RU/EN/ES/PT/FR/AR)
TimesFM 3.0 Cuts University Building Energy Forecast Error by 43%
Research·arXiv cs.LG

TimesFM 3.0 Cuts University Building Energy Forecast Error by 43%

Oct 12, 2026·2 min
Wasserstein BaryBooster Turns Diffusion Realism into Event Forecast Accuracy
Research

Wasserstein BaryBooster Turns Diffusion Realism into Event Forecast Accuracy

Oct 12, 2026·2 min
DiffusionART speeds up RL fine-tuning of video models 12x, with no EMA tuning
LLM

DiffusionART speeds up RL fine-tuning of video models 12x, with no EMA tuning

Oct 12, 2026·2 min
A Physical Transformer Built Entirely From Transistors Learns In-Context Learning in Simulation
Research

A Physical Transformer Built Entirely From Transistors Learns In-Context Learning in Simulation

Oct 12, 2026·2 min
Theory Shows Masked Diffusion on Transformers Learns from Õ(LK/ε) Samples
Research·arXiv cs.LG

Theory Shows Masked Diffusion on Transformers Learns from Õ(LK/ε) Samples

A new arXiv paper gives end-to-end guarantees for masked diffusion models: sample complexity scales as Õ(L·K_par/ε), not exponentially as V^L in sequence length.

Oct 12, 2026·2 min
SAZO speeds up LLM fine-tuning without backprop: up to 6× fewer queries to reach the same loss
Research·arXiv cs.LG

SAZO speeds up LLM fine-tuning without backprop: up to 6× fewer queries to reach the same loss

A new zeroth-order method, SAZO, learns to tell informative training examples from uninformative ones and reaches the same loss with up to 6× fewer queries than…

Oct 12, 2026·2 min
Holomorphic Neural Networks With Singular Functions Beat All Methods on Laplace Benchmarks
Research·arXiv cs.LG

Holomorphic Neural Networks With Singular Functions Beat All Methods on Laplace Benchmarks

A new architecture satisfies the Laplace equation by construction and captures corner singularities. On a CdZnTe detector, it is 1.7–2.1 times more accurate than…

Oct 12, 2026·2 min
Cox Regression Learns From Truncated Data Orders of Magnitude Faster Than EM
Research·arXiv cs.LG

Cox Regression Learns From Truncated Data Orders of Magnitude Faster Than EM

An arXiv paper proves that Cox regression can be trained efficiently on truncated and censored data with gradient methods. The speedup over EM is orders of…

Oct 12, 2026·2 min
Top themes
01CacheScope speeds up diffusion models by up to 3.57x and cuts LPIPS by 70%Oct 12, 2026·2 min02AI Error in Estimating Hidden Solar Generation Rises from 0.104 to 0.185 kWOct 12, 2026·2 min03CircuitTree assembles 39 HuBERT phoneme circuits into a syllable transcriber with 64.8% accuracyOct 12, 2026·2 min04ResonatorLM Taught to Read Memory by Token: Perplexity at 1024 Drops from 12.81 to 3.76Oct 12, 2026·2 min05Kairos: a byte-level model for all modalities converges on 440 bytesOct 12, 2026·2 min
Open LLMs Explain Satire More Cheaply Than Frontier Models, but Handle Australian Satire Worse Than American
Research·arXiv cs.CL

Open LLMs Explain Satire More Cheaply Than Frontier Models, but Handle Australian Satire Worse Than American

Oct 12, 2026·2 min
COFFEQA Adds 8.77 EM on TriviaQA to Few-Shot Distillation Using Counterfactuals
Research·arXiv cs.CL

COFFEQA Adds 8.77 EM on TriviaQA to Few-Shot Distillation Using Counterfactuals

Oct 12, 2026·2 min
MADS Trains Dialog Agents Not to Break: 1,095 Conversations, 66% With "Bad" Scenarios
Research·arXiv cs.CL

MADS Trains Dialog Agents Not to Break: 1,095 Conversations, 66% With "Bad" Scenarios

Oct 12, 2026·2 min
U4RASD Takes 1st Place at KnowledgeGraphEval-2026: 94.06 micro-F1 on Arabic
Research·arXiv cs.CL

U4RASD Takes 1st Place at KnowledgeGraphEval-2026: 94.06 micro-F1 on Arabic

Oct 12, 2026·2 min
AXLE Method Primes Annotators' Own Experience Before Hate Speech Labeling: 104 People, 1,040 Ratings
Research·arXiv cs.CL

AXLE Method Primes Annotators' Own Experience Before Hate Speech Labeling: 104 People, 1,040 Ratings

AXLE, a new pre-annotation step, asks annotators to describe personal experience before rating toxic text. In a test with 104 people, agreement dropped and label…

Oct 12, 2026·2 min
perturber: a service with 68 text perturbations tests LLM robustness
Research·arXiv cs.CL

perturber: a service with 68 text perturbations tests LLM robustness

Researchers have released perturber, a service and Python library that apply one meaning-preserving perturbation at a time from a catalog of 68 variants across 11…

Oct 12, 2026·2 min
GeopoliticalQA: bge-m3 Beats Commercial Embeddings on 5 525 Report-Based Questions
Research·arXiv cs.CL

GeopoliticalQA: bge-m3 Beats Commercial Embeddings on 5 525 Report-Based Questions

A new benchmark tests RAG on 415 geopolitical reports and 5 525 questions: compact bge-m3 outperforms commercial APIs, while generation struggles with inference and risk-assessment questions.

Oct 12, 2026·2 min
Qwen Picks What to Remember: Replay From a Frozen Model Reduces Forgetting
Research·arXiv cs.CL

Qwen Picks What to Remember: Replay From a Frozen Model Reduces Forgetting

Oct 12, 2026·3 min
CLUE Builds an Event Ontology Without Experts in Three Steps: Embeddings, Clusters, Names
Research·arXiv cs.CL

CLUE Builds an Event Ontology Without Experts in Three Steps: Embeddings, Clusters, Names

Oct 12, 2026·2 min
AutoIndexer Teaches a Language Model to Edit Its Own Output with Three Types of Edits
Research·arXiv cs.CL

AutoIndexer Teaches a Language Model to Edit Its Own Output with Three Types of Edits

Oct 12, 2026·2 min
OGPO Teaches VLA Models to Get Back on Track After a Failure, Without Online Fine-Tuning
Robotics·arXiv cs.AI

OGPO Teaches VLA Models to Get Back on Track After a Failure, Without Online Fine-Tuning

Oct 12, 2026·2 min
Seq-Drifting Generates Text in a Single Pass: Perplexity of 36.01 on LM1B
Research·arXiv cs.AI

Seq-Drifting Generates Text in a Single Pass: Perplexity of 36.01 on LM1B

Oct 12, 2026·2 min
CToM teaches LLMs to predict user replies and lifts accuracy to 79.6%
Research

CToM teaches LLMs to predict user replies and lifts accuracy to 79.6%

Oct 12, 2026·2 min
GeoSpatialBench Tests MLLMs on 28 Types of Spatial Relations Across Six Cities
Robotics

GeoSpatialBench Tests MLLMs on 28 Types of Spatial Relations Across Six Cities

Oct 12, 2026·2 min
RGR Protects LLMs from Skill Loss in Self-Distillation by Preserving Representation Geometry
Research

RGR Protects LLMs from Skill Loss in Self-Distillation by Preserving Representation Geometry

Oct 12, 2026·2 min
Researchers Propose a 3-Criterion Formula for Creativity and Test It on Chess and Picasso
Research·arXiv cs.AI

Researchers Propose a 3-Criterion Formula for Creativity and Test It on Chess and Picasso

An arXiv paper proposes a formal definition of creativity built on novelty, value and surprise, and tests it on chess moves and paintings.

Oct 12, 2026·2 min
Neural Networks Are More Modular Than Their Neurons Suggest: Three Levels of Testing
Research·arXiv cs.AI

Neural Networks Are More Modular Than Their Neurons Suggest: Three Levels of Testing

A new arXiv paper proposes three nested levels of modularity and finds that networks without neuron-level modules often have clear functional modules in subspaces.

Oct 12, 2026·2 min