Tensor Learn · field notes for model libraries
Technical answers for
local model libraries.
For the people whose disk fills with LoRAs, checkpoints, SafeTensors and local LLMs. Learn what is actually stored, what can be recovered, and where Tensor Archive helps without hand-waving.
Choose the right format
One question. The right depth.
The Wiki answers a concrete operational question. The Blog unpacks a decision or workflow. News records changes we can substantiate, not recycled AI headlines.
Where did the models go?
Paths, caches, file formats and the distinctions that keep a cleanup from becoming a recovery incident.
Tensor BlogMake the storage decision before the deletion.
Practical guides for LoRA libraries, related model families, exact recovery and fine-tuning artifacts.
Tensor NewsOnly updates worth adding to your workflow.
Product changes and carefully sourced ecosystem developments that affect local model libraries.
Start here
The first useful routes.
These are not generic explainers. Each page starts with the actual question, names the trade-off, and ends with a safe next action.
- LoRAs are taking over the SSDRecover space without pretending that quantization is a backup.
- Find the Hugging Face cacheInspect the active cache location before deleting a snapshot.
- Keep the right fine-tuning artifactsArchive completed versions; do not confuse storage with training recovery.
- One base, several adaptersRecognize repeated self-contained deployments before they fill the SSD.
- What does SafeTensors actually mean?Understand the format without confusing it with quantization or an archive.
- Find LM Studio models without guessingUse its own inventory and import workflow before you move files.