Pre-training
What it means: Pre-training is the first major phase of model development. It’s when a model is fed massive amounts of data (like books, websites, and code) so it can learn general patterns in language, images, or other types of input. What to know: Pre-training doesn’t load the model with facts. It teaches how language or images behave, so the model can later predict and generate content. How it shows up in Gloo: The models you call through Gloo AI are pre-trained by their providers, such as OpenAI, Anthropic, Google, and open-weight model developers.Post-training
What it means: Post-training covers the steps after pre-training, such as refining the model for safety, quality, or specific uses. Why it matters: Post-training turns a general-purpose model into a useful product. It might involve adding safety rules, formatting guidelines, or behavior alignment.Fine-tuning
What it means: Fine-tuning is further training of a pre-trained model on a smaller, curated dataset, so it specializes in a topic, task, or tone. Use case: If you wanted a model to work only with legal documents or theological texts, you’d fine-tune it on relevant examples to get better performance in that domain.Instruction Tuning
What it means: Instruction tuning teaches a model to follow instructions more reliably. It trains on examples of clear instructions paired with good responses. Why it helps: This is what makes models like ChatGPT feel responsive and helpful. It teaches models not just to predict, but to follow your intent.Reinforcement Learning from Human Feedback (RLHF)
What it means: In RLHF, people rate different model outputs, and the model learns to favor the kinds of responses people prefer. Analogy: It’s like giving the model a thumbs up or down for each answer, so it learns what replies people want.Alignment
What it means: Alignment means making sure a model’s behavior stays consistent with human values, safety guidelines, and ethical boundaries. Why it matters: A powerful model that isn’t aligned could give unsafe, biased, or manipulative responses. Alignment helps ensure models behave as intended even in unexpected situations. How it shows up in Gloo: Alignment is central to Gloo AI. Guarded endpoints add guardrails, output moderation, and values-aligned responses through thetradition setting. Grounded endpoints answer from content you’ve uploaded to the Data Engine, so answers draw on material you chose.
Go deeper: Guardrails and Values explains values alignment and how Gloo measures it.
Distillation
What it means: Distillation compresses a large model into a smaller, faster version while keeping most of the same abilities. Use case: Companies often use distillation to deploy smaller versions of powerful models on phones, websites, or embedded devices where resources are limited.Transfer Learning
What it means: Transfer learning means applying knowledge learned in one context to another. A model trained on general language can adapt to specialized tasks without starting from scratch. Example: You could take a model trained on internet text and quickly adapt it to answer medical questions without retraining the whole system. How it shows up in Gloo: Instead of retraining a model on your content, Gloo AI’s grounded endpoints retrieve relevant passages from content you’ve uploaded to the Data Engine and give them to the model with each request. This is called retrieval-augmented generation (RAG). You get answers that use your material, and updates to your content take effect without retraining.What’s next
To see how grounding works in an app, read Grounded Completions, follow the RAG pipeline tutorial, and see Building Trustworthy Grounded Applications.How We Talk to Models
How do we give AI models instructions, and how do they turn those prompts into useful answers?

