LLM decision trees: from Tree Prompting to agentic classification
Learned programs and promptingHow Tree Prompting learns decision trees of prompted language-model calls, and how it relates to Aug-Tree, GPTree and Agentic Classification Tree.
↗Methods & connections
How models choose,
generate, and learn.
Seven guides to language-model decisions: prompted trees, sequence energies, probabilistic attention, structured decoding, text evaluation, learned rendering, and live training.
Start with a concrete example. Follow the method to its original paper and its connections to current research.
Find the idea behind the method.
Seven guides · Original papers & current connections
Energy-based text generation: residual models, resampling and diffusion
Global sequence modelsResidual Energy-Based Models for Text Generation connects sequence-level energy, noise-contrastive estimation and importance resampling to energy-based diffusion language models.
↗Structured Decoding: Higher-Order CRFs, Diffusion Language Models, and Candidate Lattices
Joint models for parallel generationHow Cascaded Text Generation and Sequence-to-Lattice connect higher-order CRFs, joint diffusion decoding, and candidate lattices for speculative generation.
↗Attention as a Latent Variable: Modeling a Distribution over Attention
Probabilistic attention and latent inferenceA guide to Latent Alignment and Variational Attention: distributions over attention, categorical and Dirichlet alignments, marginalization, and posterior inference.
↗Evaluating the structure of long-form text
Coherence, coreference and narrative evaluationHow latent-space model criticism evaluates discourse structure, entity behavior, and topicality beyond word-level fluency in generated text.
↗Learning to render LaTeX, SMILES, HTML and music notation
Markup-to-image diffusionDiffusion models as learned renderers for mathematical formulas, molecules, table layouts, and music notation, with scheduled sampling to address exposure bias.
↗Controlling a running training job with humans and AI agents
Feedback-driven neural network optimizationInteractive Training connects live metrics to human and LLM interventions in running neural-network jobs; v2 adds typed controls and an ordered action journal.
↗Follow a research question
From the original formulation to today’s systems.
Compare what is modeled, how it is learned, and how predictions are selected. The guides make direct methodological links explicit and keep distinct approaches separate.
Explore joint decoding →Use the references
A bibliography you can take with you.
Full author lists, paper records, source links, BibTeX, CSL JSON, and Markdown come from the same bibliography.
Citations & downloads →