Nodes/Ray's Orenchi/🧰 Ray's Model Tools: Convert Model
ComfyUI Node

🧰 Ray's Model Tools: Convert Model

Convert a model file or the effective state of a connected MODEL. Inference outputs are loaded automatically when used. Non-Linear layers stay bf16 for quantized output.

By MutantSparrowΒ·Created 4 months agoΒ·Updated 3 days agoΒ· 1
🧰 Ray's Model Tools: Convert Model
  • model
  • clip
  • vae
  • model_data
  • model
  • clip
  • vae
  • model_data
  • metadata
  • report
β—„weight_formatbf16β–Ί
β—„componentsdiffusionβ–Ί
β—„pathβ–Ί
CategoryπŸ‘‘ Ray/🧰 Model Tools

Inputs (7)

NameTypeDefaultDescription
weight_formatCOMBObf16Storage precision or quantization for the result. Auto/Preserve keeps the source representation where supported. Converting to higher precision cannot recover information already lost.
componentsCOMBOdiffusionChoose diffusion weights, diffusion plus text encoders, text encoders only, or all supported components. Unselected components follow the operation's preservation rules.
modeloptMODELConnected model to process, including its active patches and LoRAs. Takes precedence over a model-data handle or file source.
clipoptCLIPOptional text encoder to include with the connected model. It must match the model architecture and selected component scope.
vaeoptVAEOptional VAE to preserve with the connected model when the operation supports complete checkpoints.
model_dataoptRAY_MODEL_DATAFile or in-memory model result from another Model Tools node. Supports data-only processing without initializing inference components.
pathoptSTRINGExisting source file or supported sharded model folder on the ComfyUI server. A connected live model or model-data handle takes precedence where available.

Outputs (6)

NameTypeDescription
modelMODELNative ComfyUI diffusion model. Inference components are initialized when this output is used; data-only workflows can use model_data instead.
clipCLIPNative text encoder when available in the source and enabled by the selected loading/component options; otherwise absent.
vaeVAENative VAE when available in the source and enabled by the selected loading/component options; otherwise absent.
model_dataRAY_MODEL_DATAModel Tools data handle for a source file or in-memory result. Connect it directly to compatible merge, conversion, inspection, or save nodes without an extra inference load.
metadataRAY_MODEL_METADATAStructured metadata for the source or result, including operation information where applicable.
reportSTRINGReadable inspection or operation report, including compatibility, format, processing, and saved-result information relevant to this operation.