# LLM Engineering Course > Understand, build, train, evaluate, align, and serve language models ## Docs - [LLM Engineering Course](https://llm.onlinekurs.training/) - [Setup — from a fresh computer to a working installation](https://llm.onlinekurs.training/setup) - [Your route through the course](https://llm.onlinekurs.training/learning-guide) ## Course - [Unit 0 — First Tiny-LM Run](https://llm.onlinekurs.training/unit-00) - [Unit 0 troubleshooting](https://llm.onlinekurs.training/unit-00-troubleshooting) - [Diagnostic and learning path](https://llm.onlinekurs.training/diagnostic) - [B01 — Godot-RL neural-foundations bridge](https://llm.onlinekurs.training/foundations-godot-rl-bridge) - [F01 — Python and NumPy](https://llm.onlinekurs.training/foundations-01-python-numpy) - [F02 — Tensors and shapes](https://llm.onlinekurs.training/foundations-02-shapes) - [F03 — Probability, logits, and loss](https://llm.onlinekurs.training/foundations-03-probability) - [F04 — Neuron and gradient descent](https://llm.onlinekurs.training/foundations-04-neuron) - [F05 — MLP, backpropagation, and tiny autograd](https://llm.onlinekurs.training/foundations-05-mlp-autograd) - [F06 — PyTorch translation and foundations checkpoint](https://llm.onlinekurs.training/foundations-06-pytorch-checkpoint) - [T01 — Bigram language-model baseline](https://llm.onlinekurs.training/bigram-baseline) - [T02 — Byte-level BPE tokenizer from scratch](https://llm.onlinekurs.training/bpe-tokenizer) - [T03 — Data pipeline & embeddings](https://llm.onlinekurs.training/data-pipeline) - [T04 — Self-attention as a number experiment](https://llm.onlinekurs.training/attention) - [T05 — Mini-GPT: assemble a tiny decoder](https://llm.onlinekurs.training/mini-gpt) - [T06 — Core gate: code, diagnose, explain](https://llm.onlinekurs.training/mini-gpt-gate) - [I01 — Decoding, sampling and the KV cache](https://llm.onlinekurs.training/inference-decoding) - [E01 — Reproducible pretraining](https://llm.onlinekurs.training/pretraining) - [E02 — Data quality and provenance](https://llm.onlinekurs.training/data-quality) - [E03 — Evaluation: make a claim the evidence can support](https://llm.onlinekurs.training/evaluation) - [E04 — Profiling: connect the model to a resource budget](https://llm.onlinekurs.training/profiling) - [C01 — Turn a company problem into a model decision](https://llm.onlinekurs.training/company-strategy) - [C02 — Auditable company data](https://llm.onlinekurs.training/company-data) - [C03 — Protected evaluation](https://llm.onlinekurs.training/company-evaluation) - [C04 — Continued pretraining without losing the baseline](https://llm.onlinekurs.training/continued-pretraining) - [A01 — SFT and LoRA: teach a response, preserve a base](https://llm.onlinekurs.training/sft-lora) - [A02 — Choose an adaptation with evidence](https://llm.onlinekurs.training/adaptation-comparison) - [A03 — DPO and the small RLHF bridge](https://llm.onlinekurs.training/dpo) - [A04 — RLHF bridge: inspect the failure mode](https://llm.onlinekurs.training/rlhf-bridge) - [Capstone — From requirement to a reversible company model](https://llm.onlinekurs.training/company-capstone) ## Applied: RAG - [R01 — From document to answer: build RAG yourself](https://llm.onlinekurs.training/rag-build) - [R02 — So it does not lie: quality, guardrails, operations](https://llm.onlinekurs.training/rag-quality) ## Applied: Agents - [AG01 — The agent loop, by hand](https://llm.onlinekurs.training/agent-loop) - [AG02 — Native tool use](https://llm.onlinekurs.training/agent-native) - [AG03 — Prompt injection: measure what each defence buys](https://llm.onlinekurs.training/agent-injection) ## Reference - [What's new](https://llm.onlinekurs.training/changelog) - [Glossary](https://llm.onlinekurs.training/glossary) - [Hardware, time, and cloud safety](https://llm.onlinekurs.training/hardware) - [Course data policy](https://llm.onlinekurs.training/data-policy) ## Deutsch ### Docs - [LLM-Engineering-Kurs](https://llm.onlinekurs.training/de) - [Einrichtung — vom frischen Computer zur funktionierenden Installation](https://llm.onlinekurs.training/de/setup) - [Dein Weg durch den Kurs](https://llm.onlinekurs.training/de/learning-guide) ### Kurs - [Einheit 0 — Erster Tiny-LM-Lauf](https://llm.onlinekurs.training/de/unit-00) - [Fehlerbehebung für Einheit 0](https://llm.onlinekurs.training/de/unit-00-troubleshooting) - [Diagnose und Lernpfad](https://llm.onlinekurs.training/de/diagnostic) - [B01 — Godot-RL-Brücke für Neural Foundations](https://llm.onlinekurs.training/de/foundations-godot-rl-bridge) - [F01 — Python und NumPy](https://llm.onlinekurs.training/de/foundations-01-python-numpy) - [F02 — Tensoren und Shapes](https://llm.onlinekurs.training/de/foundations-02-shapes) - [F03 — Wahrscheinlichkeit, Logits und Loss](https://llm.onlinekurs.training/de/foundations-03-probability) - [F04 — Neuron und Gradient Descent](https://llm.onlinekurs.training/de/foundations-04-neuron) - [F05 — MLP, Backpropagation und kleines Autograd](https://llm.onlinekurs.training/de/foundations-05-mlp-autograd) - [F06 — PyTorch-Übersetzung und Grundlagen-Checkpoint](https://llm.onlinekurs.training/de/foundations-06-pytorch-checkpoint) - [T01 — Bigramm-Sprachmodell als Baseline](https://llm.onlinekurs.training/de/bigram-baseline) - [T02 — Byte-Level-BPE-Tokenizer von Grund auf](https://llm.onlinekurs.training/de/bpe-tokenizer) - [T03 — Datenpipeline & Embeddings](https://llm.onlinekurs.training/de/data-pipeline) - [T04 — Self-Attention als Zahlenexperiment](https://llm.onlinekurs.training/de/attention) - [T05 — Mini-GPT: einen kleinen Decoder zusammensetzen](https://llm.onlinekurs.training/de/mini-gpt) - [T06 — Kernprüfung: programmieren, Fehler finden, erklären](https://llm.onlinekurs.training/de/mini-gpt-gate) - [I01 — Decoding, Sampling und der KV-Cache](https://llm.onlinekurs.training/de/inference-decoding) - [E01 — Reproduzierbares Vortraining](https://llm.onlinekurs.training/de/pretraining) - [E02 — Datenqualität und Provenienz](https://llm.onlinekurs.training/de/data-quality) - [E03 — Evaluation: Aussagen mit überprüfbarer Evidenz](https://llm.onlinekurs.training/de/evaluation) - [E04 — Profiling: vom Modell zum Ressourcenbudget](https://llm.onlinekurs.training/de/profiling) - [C01 — Vom Firmenproblem zur Modellentscheidung](https://llm.onlinekurs.training/de/company-strategy) - [C02 — Prüffähige Firmendaten](https://llm.onlinekurs.training/de/company-data) - [C03 — Geschützte Evaluation](https://llm.onlinekurs.training/de/company-evaluation) - [C04 — Continued Pretraining ohne Verlust der Baseline](https://llm.onlinekurs.training/de/continued-pretraining) - [A01 — SFT und LoRA: Antworten lernen, Basismodell erhalten](https://llm.onlinekurs.training/de/sft-lora) - [A02 — Anpassung anhand von Evidenz wählen](https://llm.onlinekurs.training/de/adaptation-comparison) - [A03 — DPO und die kleine RLHF-Brücke](https://llm.onlinekurs.training/de/dpo) - [A04 — RLHF-Brücke: Fehler sichtbar machen](https://llm.onlinekurs.training/de/rlhf-bridge) - [Capstone — Von der Anforderung zum reversiblen Firmenmodell](https://llm.onlinekurs.training/de/company-capstone) ### Angewandt: RAG - [R01 — Vom Dokument zur Antwort: RAG selbst gebaut](https://llm.onlinekurs.training/de/rag-build) - [R02 — Damit es nicht lügt: Qualität, Guardrails, Betrieb](https://llm.onlinekurs.training/de/rag-quality) ### Angewandt: Agenten - [AG01 — Die Agentenschleife, von Hand](https://llm.onlinekurs.training/de/agent-loop) - [AG02 — Native Tool-Use](https://llm.onlinekurs.training/de/agent-native) - [AG03 — Prompt-Injection: messen, was jede Abwehr bringt](https://llm.onlinekurs.training/de/agent-injection) ### Referenz - [Neu im Kurs](https://llm.onlinekurs.training/de/changelog) - [Glossar](https://llm.onlinekurs.training/de/glossary) - [Hardware, Zeit und Cloud-Sicherheit](https://llm.onlinekurs.training/de/hardware) - [Datenrichtlinie des Kurses](https://llm.onlinekurs.training/de/data-policy) ## Agent resources - [llms-full.txt](https://llm.onlinekurs.training/llms-full.txt): The full Markdown of every page in one file. - [Page Markdown](https://llm.onlinekurs.training/index.md): Append `.md` to any page URL to fetch that page as raw Markdown. - [JSON API](https://llm.onlinekurs.training/api/docs/pages.json): Page index of the JSON docs API; each entry links the page's JSON and Markdown forms. Described by the OpenAPI document at https://llm.onlinekurs.training/openapi.json. - [API catalog](https://llm.onlinekurs.training/.well-known/api-catalog): RFC 9727 linkset of the APIs documented here. - [agent-readability.json](https://llm.onlinekurs.training/agent-readability.json): Manifest of every agent-facing artifact on this site. - [Sitemap](https://llm.onlinekurs.training/sitemap.xml): Every indexable page URL with its last-modified date.