HubClient

classin packageklyn.llm.inference.hub

public class HubClient

HubClient

Verified, resumable Hugging Face model downloads without external tooling. Resolves a branch once, downloads from its immutable commit, checks file sizes and LFS SHA-256, then publishes a complete local snapshot. No remote code is run.

Example

snapshot = HubClient.download("Qwen/Qwen3.5-9B")

Methods

Modifier and Type	Member	Description
public static	availableQuantizations `availableQuantizations(repository as String, revision as String = "main") as ArrayList<ModelQuantization>`	Lists recognized text-model GGUF variants published by a repository, not execution backend capabilities.
public static	download `download(repository as String, revision as String = "main", cache as Path = null, offline as Boolean = false, token as String = "", verbose as Boolean = true) as HubSnapshot`	Downloads or reuses a complete, pinned local snapshot.
public static	downloadGGUF `downloadGGUF(repository as String, quantization as ModelQuantization, revision as String = "main", offline as Boolean = false, token as String = "", verbose as Boolean = true) as HubSnapshot`	Downloads only the selected GGUF file group, using pinned revision URLs, bounded resumable transfers, byte progress and SHA-256 verification.