Memuat…

Course Forum Asisten Sumber Cara belajar RAG Peta modul

Masuk Mulai →

socrates

Semua tentang AI, satu jalur belajar. Gratis, bersumber terbuka, dan setiap jawaban punya rujukan.

Navigasi

Course
Forum
Asisten
Sumber
Cara belajar
RAG
Peta modul

Dibangun dengan

Next.js · Tailwind ·
Lenis + GSAP · SQLite + BM25
(RAG hybrid: cosine + BM25, RRF)

© 2026 socratesgratis · terbuka · sumber selalu dicantumkan

AI Fundamentals — socrates · socrates

AI Fundamentals

Dua belas babak. Nol sampai agent.

M0–M11: dari nol sampai agent — teori digarap lewat source, praktik lewat project.

Kurikulum Inti Project Lab

Modul: 12
Lesson: 63
Project: 13
Durasi: ±25.5 minggu

Progress Kurikulum Inti0/63 lesson

Fondasi & Lingkungan

mental model AI, terminologi dasar, Python, Git, dan setup toolchain

Buka bab lengkap →

AI / ML / DL / Generative AI / LLM: bedanya AI vs Machine Learning (video)AI, ML, DL & Generative AI dijelaskan (video)Peta jalan roadmap.sh Python + venv dalam satu sesi +1 lesson lainnya →

1 mgg0/6 selesai

Matematika Inti untuk AI

Linear algebra, kalkulus, proba & statistik terapan

Buka bab lengkap →

Tensor & vector space Essence of Linear Algebra Turunan, chain rule, gradien Probabilitas untuk ML: Bayes, Gaussian & MLE Project: Gradient descent from scratch

1.5 mgg0/5 selesai

Machine Learning Engineering

Model klasik, evaluasi, feature engineering, MLOps

Buka bab lengkap →

Data pipeline & split train-valid-test Supervised: linier sampai ensemble Precision, recall, F1, ROC A Few Useful Things about ML (paper)MLOps dasar: tracking eksperimen +1 lesson lainnya →

3 mgg0/6 selesai

Deep Learning

NN dari nol, PyTorch, CNN, optimizer

Buka bab lengkap →

Perceptron sampai MLP Backpropagation dipahami dari nol PyTorch 60-menit: tensor & autograd CNN & computer vision Dropout, weight decay, optimizer +1 lesson lainnya →

3 mgg0/6 selesai

Large Language Models

Tokenisasi, attention, transformer, ekosistem HF

Buka bab lengkap →

Tokenisasi BPE/WordPiece Attention Is All You Need (paper)Let's build GPT dari nol Embedding & sampling (temp, top-p)Project: mini-GPT lokal

2.5 mgg0/5 selesai

Prompt Engineering

Few-shot, CoT, structured output, guardrails

Buka bab lengkap →

Zero-shot → few-shot Chain-of-thought (paper)Structured output JSON + schema Prompt injection & mitigasinya Project: Prompt Bench

1.5 mgg0/5 selesai

Context Engineering

Context window, memory, kompresi, caching

Buka bab lengkap →

Context window & Lost in the Middle Memory: windowing, summary, compression Context caching & biaya token Tool schema sebagai format konteks Project: Context Bucket

1.5 mgg0/5 selesai

Retrieval-Augmented Generation

Chunking, retrieval, rerank, evaluasi, RAG lanjutan

Buka bab lengkap →

RAG paper asli (2005.11401)Chunking: size, overlap, section-aware Embedding & pgvector praktis Hybrid retrieval + reranker Evaluasi RAG dengan RAGAS +1 lesson lainnya →

2.5 mgg0/6 selesai

AI Agents & Tool Use

Agent loop, ReAct, function calling, MCP

Buka bab lengkap →

ReAct: reason + act (paper)Agent loop & tool calling Multi-agent & orchestrator pattern MCP: Model Context Protocol Project: Mini Agent

2 mgg0/5 selesai

Fine-tuning & Alignment

LoRA/QLoRA, dataset, RLHF/DPO, evals

Buka bab lengkap →

LoRA & QLoRA (paper)Format dataset instruct & hygiene SFT praktis dengan TRL/Unsloth DPO: alignment tanpa reward model Project: Finetune Mini + LLM-judge

2 mgg0/5 selesai

AI Engineering di Production

Serving, biaya/latensi, guardrails, observability

Buka bab lengkap →

Serving: vLLM, Ollama, endpoint Prompt & semantic caching Token, cost & latency observability Evals sebagai CI Project: ship RAG ke production

2 mgg0/5 selesai

Capstone — AI Study Buddy

Produk RAG/agent end-to-end, ingest banyak sumber

Buka bab lengkap →

Scope & spesifikasi capstone Ingest 20 sumber nyata Golden set: >=20 QA, pass >=80%Deploy + demo + executive summary

3 mgg0/4 selesai

Mode lesson

TEXT: — baca terstruktur
VIDEO: — playlist kurasi
ARTICLE: — paper & medium
PROJECT: — langsung bangun

PROJECT = learning-by-doing: kerjakan, commit, dan tandai selesai.

Mulai dari nol — tanpa daftar, tanpa paywall

Yang pertama: pelajari cara belajarnya →

Halaman lesson per-babak + tracking progress.