Tensor Learn launches with the questions people ask before reclaiming model storage

Tensor Learn is our public reference layer for people who do not need another generic AI explainer. They need to know where the models are, what the files mean, what a storage result actually proves, and how to recover before cleanup.

Why this exists

Local AI model libraries accumulate in non-obvious places: a runtime folder, a hub cache, a project output directory, a collection of LoRAs with a half-remembered base model. Search results often jump from that real operational problem to a drastic action—delete, quantize, convert—without distinguishing the requirement.

Tensor Archive has a specific role in that decision: archive related local model versions, verify recoverability and restore a separate exact copy before a person reclaims a redundant working copy. It is not the right answer to every storage question, so the content needs to make the boundary explicit.

Three lanes, not one content bucket

Tensor Wiki answers direct operational questions: where a cache lives, which setting changes a path, or what one technical term means. Statements that are version-sensitive link to the original documentation.

Tensor Blog handles a decision or workflow: a full LoRA library, an exact-storage-versus-quantization choice, or the completed artifacts from a fine-tuning project. These guides lead with the problem and the limitation, then describe the relevant product workflow.

Tensor News records product and ecosystem changes only when they affect a real local storage or recovery decision. It is intentionally not a broad AI news feed.

The editorial standard

• Answer the question before introducing the product.
• Put the workload and limitation next to every benchmark number.
• State when Tensor Archive is not the right tool.
• Use technical terms precisely, but explain them without talking down to readers.
• Never tell someone to remove original model data before they have verified and tested a separate restore.

The first useful pages

Start in the Wiki if you need an answer now: Hugging Face cache location, Ollama model paths or LoRA versus QLoRA. Go to the Blog when the question is whether to archive, delete, quantize or preserve a related family. We will add pages when they meet the same standard, not just because a keyword exists.