Evaluation & Observability
Curated Evaluation & Observability tutorials.
AI Anomaly Detection for Time Series: From Statistical to Deep Learning Approaches
Build production anomaly detection systems for time series data using statistical methods, isolation forest, LSTM autoencoders, and modern time series foundation models for infrastructure and IoT monitoring.
IntermediateBuilding Enterprise-Grade RAG 2.0 Systems: A Complete Practice from Document Parsing to Knowledge Retrieval
This article systematically introduces the construction and optimization methods of enterprise-grade RAG 2.0 systems, covering key technologies such as document parsing, query rewriting, hybrid retrieval, ranking fusion, ontology constraints, and cache optimization. Combined with real-world scenarios in manufacturing and finance, it explains in detail how to address core challenges like parsing complex document structures, multi-turn dialogue anaphora resolution, and balancing retrieval precision and recall. It also introduces ontology-driven semantic constraints and caching mechanisms to improve accuracy and response efficiency in professional domains. Suitable for developers with basic RAG knowledge who want to build production-level systems.
IntermediateLangSmith for LLM Evaluation: Building Systematic Feedback Loops
LangSmith LLM Evaluation Workflow (2026): Trace → Dataset → Evaluator (including LLM-as-judge) → Experiment — the four-piece suite that turns "feels better" into measurable progress. Includes @traceable code, weekly evaluation loops, bias calibration for LLM judges, and comparison vs Langfuse.
IntermediateLarge Model Post-Training in Practice: From SFT to RL — The Complete Tech Stack
This article systematically explains the key techniques of large model post-training, including supervised fine-tuning (SFT), parameter-efficient fine-tuning (PEFT), reinforcement learning from human feedback (RLHF), and on-policy distillation (OPD). It focuses on the principles, pros and cons, and applicable scenarios of each method, and introduces a stability-plasticity trade-off framework to quantify the general capability loss caused by fine-tuning. By comparing the forgetting characteristics of full fine-tuning, LoRA, OFT, and other PEFT methods, it reveals that the destruction of activation space geometric structure is the key mechanism of forgetting. Finally, it summarizes the advantages of OPD as a new paradigm and provides practical guidelines and FAQs.
IntermediatePrometheus + Grafana for AI Applications: Monitoring AI services Guide 2026
Prometheus + Grafana for AI Applications: monitoring AI services 2026 Introduction Set up comprehensive monitoring for LLM API costs, latency, and error rates. This guide shows you how to effectively use Prometheus + Grafana in your AI development
IntermediateRAGAS Evaluation: Developer Guide and Quick Start 2026
RAGAS Evaluation: Developer Guide 2026 What is RAGAS Evaluation? **RAGAS Evaluation** enables evaluate RAG systems quantitatively. This guide covers everything you need to get started quickly. Why Use RAGAS Evaluation? - Solves the specific probl
IntermediateWhyLabs AI Observatory: Complete Setup Guide
WhyLabs and Profile-Based ML Observability (2026): Monitor statistical profiles of data instead of raw data—whylogs is open-source, KB-scale summaries, raw data never leaves the boundary, inherently compliant. Predict drift without labels, extend to the LLM era (text metrics + embedding space drift), and complement trace-level observability.