H
Howardism
Howardism · Vol. 03Plate I · No. 01

LLM Architecture, tagged.

Notes44TagLLM ArchitectureOldest10 Apr 2026Newest23 Sept 2026

Every article tagged llm architecture, newest first.

  1. C01
    Native Multimodal Modeling: Fusion Depth and I/O Duality
    LLM ArchitectureMultimodalTraining
    · 27′
  2. C02
    Shared-Budget Compute Allocation
    Test Time ComputeReasoning ModelsMetacognition+2
    · 21′
  3. C03
    Inference-Time Architecture Search
    Test Time ComputeInference ScalingLLM Architecture+2
    · 12′
  4. C04
    Rationale Bootstrapping (STaR)
    LLM ArchitecturePost TrainingReasoning+2
    · 26′
  5. C05
    Live-Path Minimalism
    LLM ArchitectureInferenceMultimodal
    · 21′
  6. C06
    Context Lifecycle Management
    Context ManagementAgent EngineeringLLM Architecture+1
    · 70′
  7. C07
    Knowledge-Centric Self-Improvement
    Agent EngineeringKnowledge ManagementRecursive Self Improvement+2
    · 47′
  8. C08
    Tool-Output Pruning
    Context ManagementAgent EngineeringCost+1
    · 41′
  9. C09
    Benchmark Contamination and Decontamination
    LLM ArchitectureCapability EvaluationBenchmarks+4
    · 29′
  10. C10
    Benchmark Score Redundancy
    LLM ArchitectureCapability EvaluationBenchmarks+3
    · 47′
  11. C11
    Measuring Beyond Accuracy Saturation
    LLM ArchitectureCapability EvaluationBenchmarks+3
    · 61′
  12. C12
    Asynchronous RL for LLMs
    LLM ArchitectureReinforcement LearningPost Training+2
    · 34′
  13. C13
    Group Relative Policy Optimization (GRPO)
    LLM ArchitectureReinforcement LearningPost Training+1
    · 34′
  14. C14
    Single-Rollout Optimization
    LLM ArchitectureReinforcement LearningPost Training+2
    · 19′
  15. C15
    Compute-Controlled Benchmarking
    LLM ArchitectureCapability EvaluationBenchmarks+2
    · 78′
  16. C16
    Inference Efficiency as Capability
    LLM ArchitectureInference ScalingTest Time Compute+2
    · 50′
  17. C17
    Large-Scale Test-Time Compute
    LLM ArchitectureTest Time ComputeInference Scaling+2
    · 41′
  18. C18
    Latent Capability Overhang
    LLM ArchitectureTest Time ComputeCapability Trajectory+1
    · 31′
  19. C19
    The Open-Weight Frontier Gap
    LLM ArchitectureOpen WeightsCapability Evaluation+1
    · 36′
  20. C20
    Repository Exploration Subagent
    Agent EngineeringLLM ArchitectureContext Management+1
    · 15′
  21. C21
    The Abstraction Barrier
    LLM ArchitectureTheoryLimits+2
    · 6′
  22. C22
    Advantages of Digital Intelligence
    LLM ArchitectureTheoryDigital Intelligence+1
    · 4′
  23. C23
    Artificial Superintelligence (ASI)
    LLM ArchitectureTheorySuperintelligence+1
    · 14′
  24. C24
    Fundamental Limits of ASI
    LLM ArchitectureTheoryLimits+1
    · 4′
  25. C25
    Instrumental Convergence
    LLM ArchitectureAlignmentSafety+2
    · 12′
  26. C26
    Transformative Creativity
    LLM ArchitectureTheoryCreativity+2
    · 24′
  27. C27
    Universal AI (AIXI)
    LLM ArchitectureTheoryUniversal AI+2
    · 6′
  28. C28
    Task Time-Horizon Scaling
    LLM ArchitectureCapability EvaluationBenchmarks+1
    · 28′
  29. C29
    Agent-Native Infrastructure
    Agent EngineeringLLM Architecture
    · 9′
  30. C30
    Agentic Loops Overtake Bespoke Systems
    Agent EngineeringAI For MathematicsLLM Architecture
    · 39′
  31. C31
    Jagged Intelligence (Ghosts, Not Animals)
    LLM ArchitectureAI SafetyMental Model
    · 14′
  32. C32
    Software 3.0
    LLM ArchitectureSoftware Paradigm
    · 5′
  33. C33
    The Verifiability Thesis
    LLM ArchitectureLLM EvaluationAgent Engineering
    · 21′
  34. C34
    Encoder-Free Early Fusion
    LLM ArchitectureMultimodal
    · 25′
  35. C35
    Interaction / Background Model Split
    LLM ArchitectureAgent EngineeringMultimodal
    · 32′
  36. C36
    Interaction Models
    LLM ArchitectureMultimodalHuman AI Collaboration
    · 27′
  37. C37
    The Bitter Lesson
    LLM ArchitecturePrinciple
    · 11′
  38. C38
    Time-Aligned Micro-Turns
    LLM ArchitectureMultimodalInference
    · 11′
  39. C39
    Turn-Based Interface Bottleneck
    Human AI CollaborationLLM ArchitectureInterface
    · 8′
  40. C40
    Context Window Smart Zone
    LLM ArchitectureAgent EngineeringContext Management
    · 20′
  41. C41
    Harness Shrinkage as Models Improve
    LLM ArchitectureAgent EngineeringHarness
    · 51′
  42. C42
    Client-Side Agent Optimization
    Agent EngineeringLLM ArchitectureOptimization+1
    · 21′
  43. C43
    Agent Harness Engineering
    Agent EngineeringLLM ArchitectureSoftware Development
    · 47′
  44. C44
    LLM-as-Compiler Knowledge Base
    Knowledge ManagementLLM ArchitecturePersonal Knowledge Base
    · 55′