shipitandpray/TurboQuant — Quantize LLM

Compress any LLM up to 6x in one step. Quantize to GGUF, GPTQ, or AWQ and optionally push to HuggingFace Hub.

View on GitHub

Trust Signals

Scorecard Score
not yet scored
Maintenance Recency
Stale
License
None
namedescriptionrequireddefault
modelHuggingFace model ID or path to local model directoryyes
formatOutput format: gguf, gptq, awq, or allnogguf
bitsQuantization bits: 2, 3, 4, 5, or 8no4
targetTarget platform: ollama, vllm, llamacpp, lmstudio (overrides format)no""
push-to-hubHuggingFace repo to upload quantized model (e.g. user/model-GGUF)no""
evalRun quality evaluation after quantizationnofalse
outputOutput directory for quantized modelno./turboquant-output
hf-tokenHuggingFace API token (required for --push-to-hub and gated models)no""
namedescription
output-dirDirectory containing quantized model files
reportPath to the JSON compression report