# Artificial Intelligence

Frontier machine learning: large language models, reasoning, alignment, interpretability, agents, generative models, efficient inference and AI for science.

## Topics

- [LLM Architectures & Pretraining](https://space-kitty.com/topics/llm-architectures) (334 papers): How frontier language models are built: transformer variants, mixture-of-experts, state-space and hybrid models, long-context attention, tokenization and pretraining recipes.
- [Scaling Laws & Compute-Optimal Training](https://space-kitty.com/topics/scaling-laws) (216 papers): Empirical laws relating loss and capability to parameters, data and compute; data-constrained and synthetic-data scaling; emergent abilities and their measurement.
- [Reasoning Models & Test-Time Compute](https://space-kitty.com/topics/reasoning-models) (256 papers): Chain-of-thought, reinforcement learning with verifiable rewards, process reward models and test-time scaling behind modern reasoning models.
- [AI Alignment & Safety](https://space-kitty.com/topics/alignment-safety) (247 papers): RLHF and its successors, constitutional and preference optimization methods, scalable oversight, dangerous-capability evaluations, reward hacking and deceptive behavior.
- [Mechanistic Interpretability](https://space-kitty.com/topics/interpretability) (259 papers): Reverse-engineering neural networks: superposition, sparse autoencoders and transcoders, circuits and attribution graphs, probing and activation steering.
- [LLM Agents & Tool Use](https://space-kitty.com/topics/ai-agents) (265 papers): Agents that plan and act: tool and function calling, computer-use and coding agents, multi-agent systems, agent benchmarks and agent protocols.
- [Generative & Multimodal Models](https://space-kitty.com/topics/generative-multimodal) (273 papers): Diffusion and flow-matching models, vision-language models, video generation and learned world models.
- [Efficient Training & Inference](https://space-kitty.com/topics/efficient-ai) (253 papers): Quantization, distillation, sparsity, speculative decoding, KV-cache compression and IO-aware kernels that make frontier models cheaper to train and serve.
- [AI for Science](https://space-kitty.com/topics/ai-for-science) (250 papers): Machine learning that accelerates discovery: protein structure, machine-learned force fields, materials discovery, weather and autonomous laboratories.
- [Robot Learning & Embodied AI](https://space-kitty.com/topics/robotics-embodied) (264 papers): Vision-language-action models, large-scale imitation learning, sim-to-real transfer, dexterous manipulation and humanoid control.

## Dossiers

- [LLM Architectures & Pretraining: State of the Field (October 2026)](https://space-kitty.com/dossiers/llm-architectures) ($0.20): How frontier language models are built and pretrained as of October 2026: sparse mixture-of-experts scaling, the competing full, sparse-compressed and hybrid linear attention designs, million-token context, optimizers, low-precision training, residual-stream redesigns, data and scaling laws. The central takeaway is that the largest open-weight models surveyed here are now trillion-parameter-scale MoEs activating roughly 3-6% of their weights per token, while the field has not converged on attention: leading labs ship mutually incompatible bets on hybrid linear recurrences, learned sparse or compressed attention, and plain full attention.
---
Space Kitty Archive · https://space-kitty.com · API: https://space-kitty.com/api/v1 · MCP: https://space-kitty.com/mcp · Premium: $0.20/request via x402 (Base Sepolia (testnet)). Metadata: arXiv (CC0), OpenAlex (CC0), Europe PMC (CC BY/CC0 works), ClinicalTrials.gov.