Learn post-training
Start with the complete guide to post-training, then go as deep as you need.
Start here
What is post-training? A complete guide for LLM teams
Post-training explained end to end: SFT, preference optimization, RL with verifiable rewards, distillation and evaluation — what each stage changes in the model, what data it needs and what it costs.
Read →GuideSFT vs DPO vs GRPO: choosing a post-training method
When supervised fine-tuning is enough, when preference optimization pays off, and when reinforcement learning with verifiable rewards is worth the compute.
Read →ComparisonFine-tuning vs RAG: a decision guide with failure modes
Fine-tuning changes how a model behaves; retrieval-augmented generation changes what it can see. A practical comparison on accuracy, freshness, cost, latency and governance, with a decision table.
Read →Decisions
Fine-tuning vs prompt engineering — and how much data fine-tuning needs
When a better prompt is enough, when fine-tuning pays off, and how many training examples you need for format, domain and reasoning changes. With a worked token-cost comparison.
Read →ComparisonH100 vs H200 vs B200 for fine-tuning
Which NVIDIA GPU to fine-tune on: published memory, bandwidth and BF16 compute for H100, H200 and B200, what each means for LoRA and full fine-tuning, and how to compare cost per run instead of cost per hour.
Read →GuideLoRA vs full fine-tuning: quality, cost and serving trade-offs
Rank, target modules, memory math and where adapters stop being enough — with real numbers from 8B to 70B.
Read →GuideChoosing a post-training platform: a buyer's guide
Hosted fine-tuning APIs, managed platforms and DIY stacks compared on control, cost, security and time-to-production.
Read →Evaluation
The release-gate checklist for fine-tuned models
A copyable checklist and YAML template for releasing a fine-tuned LLM: eval set hygiene, statistics, judge calibration, safety, operations and rollout. Built from the failures we see most.
Read →GuideEvaluating fine-tuned models: how to test your evals
Golden sets, LLM-as-judge calibration, contamination checks and the regression gates that keep a release honest.
Read →ResearchHow many eval examples do you need to trust a 3-point gain?
Power analysis for LLM evals with a reproducible simulation: minimum detectable effect by eval size, why paired comparisons need fewer items, and a sizing table you can use before you build the golden set.
Read →Cost
How much does it cost to fine-tune an LLM?
A line-item breakdown of GPU hours, data, evals and serving, with worked examples for 8B, 32B and 70B models.
Read →ResearchWhen does a fine-tuned model beat the API on cost?
A transparent cost model comparing a self-hosted fine-tuned model with a hosted frontier API: the break-even token volume, the full cost curve, and which assumptions actually move the answer.
Read →ResearchHow much GPU memory does fine-tuning need?
The memory formula for full fine-tuning, LoRA and QLoRA — weights, gradients, Adam state and activations — with a computed table for 8B to 70B models and the minimum H100 count for each.
Read →Glossary
Short, dense definitions with the equations.
Direct Preference Optimization (DPO)
A preference-learning method that trains directly on chosen/rejected response pairs, optimising the RLHF objective in closed form.
Read →GlossaryGroup Relative Policy Optimization (GRPO)
A reinforcement-learning method that scores a group of sampled completions per prompt and uses their mean reward as the baseline, removing PPO's value network.
Read →GlossaryLoRA (Low-Rank Adaptation)
A parameter-efficient fine-tuning method that freezes the base weights and learns a low-rank update ΔW = BA for selected layers.
Read →GlossaryReward hacking
When a policy learns to score highly on its reward function by exploiting gaps in it rather than by solving the intended task.
Read →GlossaryKL divergence penalty
A regulariser that penalises the trained policy for moving away from a reference model, measured by KL divergence.
Read →GlossaryLLM-as-a-judge
Using a language model, prompted with a rubric, to score or compare outputs from another model.
Read →GlossaryCatastrophic forgetting
The degradation of a model's previously learned abilities after training on new, narrower data.
Read →By industry
Post-training for customer support agents
How to fine-tune an LLM support agent that follows refund and escalation policy: data from resolved tickets, policy-aware rewards, evals by intent, and the metrics that matter in production.
Read →Use casePost-training for banks and insurers
Fine-tuning LLMs under model risk management: how SR 11-7 expectations map onto post-training, what validators ask for, and use cases from claims extraction to KYC review.
Read →Use casePost-training for healthcare and life sciences
How to fine-tune LLMs on clinical text: HIPAA Safe Harbor and Expert Determination de-identification, medical coding and documentation use cases, and evaluating clinical models.
Read →GuideHow to fine-tune an LLM on Databricks
Unity Catalog data, governed GPU clusters, MLflow lineage and Model Serving — the end-to-end pattern for post-training inside your lakehouse.
Read →Bring a model and a metric.
Tell us what the model should get better at. A post-training engineer replies within one business day with a plan, a compute estimate and an eval design.