Skip to main content

Module online

Module online 

Source
Expand description

Online learning and dynamic retraining infrastructure Online Learning Infrastructure for Dynamic Model Retraining

This module provides incremental model updates without full retraining, supporting continuous improvement in production ML systems.

§References

  • [Bottou 2010] “Large-Scale Machine Learning with Stochastic Gradient Descent”
  • [Crammer et al. 2006] “Online Passive-Aggressive Algorithms”

§Toyota Way Principles

  • Kaizen: Continuous model improvement via online learning
  • Jidoka: Drift detection stops bad predictions automatically
  • Just-in-Time: Retrain only when drift detected, not on schedule

Modules§

corpus
Corpus Management for Online Learning
cpt
Continual Pre-Training (CPT) Pipeline (GH-448)
curriculum
Curriculum Learning for Progressive Training
dam
Differentiable Adaptive Merging (DAM) (GH-446)
distillation
Knowledge Distillation for Model Compression
distillation_advanced
Advanced Knowledge Distillation Strategies (GH-451)
dpo
Direct Preference Optimization (DPO) (GH-449)
drift
Drift Detection for Triggering Model Retraining
eval_harness
Evaluation Harness for Standard Benchmarks (GH-454)
moe_construction
Mixture of Experts (MoE) Construction from Dense Models (GH-445)
orchestrator
Retrain Orchestrator for Drift-Triggered Model Updates
per_layer_merge
Per-Layer Merge Granularity for Model Merging (GH-452)
rlvr
Reinforcement Learning on Verifiable Rewards (RLVR) (GH-450)
tokenizer_surgery
Tokenizer Surgery for Vocabulary Transplantation (GH-447)

Structs§

OnlineLearnerConfig
Configuration for online learning
OnlineLinearRegression
Simple online linear regression using SGD
OnlineLogisticRegression
Simple online logistic regression using SGD

Enums§

LearningRateDecay
Learning rate decay schedules

Traits§

OnlineLearner
Online learning capability for incremental model updates
PassiveAggressive
Passive-Aggressive online learning for classification