0.0.9

ComfyUI_GGUF_windows_portable

key highlights:

• gguf-node (1.0.7) for comfyui (0.3.11)
• tensor cutter (beta) and convertor (alpha) implemented; DIY: create/make your own quantized model(s) possible
• new sampler: res_multistep added; Nvidia Cosmos 7B & 14B: text to video and image to video model(s) supported
• some cases of ancestral samplers not being deterministic fixed, etc.
• others; please refer to comfyui

0.0.8

ComfyUI_GGUF_windows_portable

key highlights:

• gguf-node (1.0.6) for comfyui (0.3.10)
• major code review and restructure; unnecessary scripts, comments and spaces removed; redirect the loading flow for better performance
• convertor (alpha) implemented; able to convert safetensors to gguf (make your own gguf possible)
• tensor cutter (beta) implemented; able to quantize safetensors from bf16 to fp8 (DIY your own quantized model)
• embedded engine gguf-connector upgraded to the latest version (1.0.8)
• others; please refer to comfyui

0.0.7

ComfyUI_GGUF_windows_portable

key highlights:

• gguf-node (1.0.4) for comfyui (0.3.10)
• tensor cutter (beta) implemented; able to quantize safetensors bf16 to fp8
• for the output above, fp8-e4m3fn scaled safetensors with around 50% decrease in file size; load faster but compensate for quality loss
• convertor (alpha) implemented; able to convert safetensors to gguf straight
• embedded engine gguf-connector upgraded to the latest version (1.0.7)
• others; please refer to comfyui

0.0.6

ComfyUI_GGUF_windows_portable

key highlights:

• gguf-node (1.0.2) for comfyui (0.3.10)
• gguf-connector deployment copy is now attached to the node directory; all the connection redirected already; and the default library in comfyui should be enough; no extra dependencies needed right away
• major code review and upgrade: all the class names were revised; old workflows might need to update (just select the new gguf loader again)
• this new update allows both comfyui-gguf and gguf node coexisting without conflicts; as they don't share the same class name anymore
• gguf convertor (alpha) implemented; able to convert safetensors to gguf
• others; please refer to comfyui

0.0.5

ComfyUI_GGUF_windows_portable

key highlights:

• gguf-node (this version will be different from the pack version; tag is the pack version) for comfyui (0.3.10)
• major upgrade; new tool gguf convertor (alpha) implemented (the last option of the dropdown menu)
• able to convert safetensors to gguf; easy to operate; just select and run; the converted gguf will be saved to output folder
• still opt to clone node and connector deployment copy via console by: py -m gguf_node
• others; please refer to comfyui

0.0.4

ComfyUI_GGUF_windows_portable

key highlights:

• gguf-node (will use the version recorded in comfy registry to avoid confusion; tag above is the pack version; and the one shown on readme.md is the pypi package version; those are three different things) for comfyui (this will be filled up by the pack version from comfyui side for better referencing)
• code updated on both node and connector
• opt to clone node and connector deployment copy via console
• test pack workflow provided, etc.
• others; please refer to comfyui

0.0.3

ComfyUI_GGUF_windows_portable

key highlights:

• gguf-node (beta) for comfyui
• legacy sdxl model supported (with workflow)
• available to clone node and connector deployment copy via console
• test pack updated; additional test pack (pony) provided, etc.
• others; please refer to comfyui

0.0.2

ComfyUI_GGUF_windows_portable

key highlights:

• gguf-node (beta) for comfyui
• aura model supported (with workflow inside)
• available to clone from pypi pack via console
• test pack updated; code polished, etc.
• others; please refer to comfyui

0.0.1

ComfyUI_GGUF_windows_portable

key highlights:

• gguf-node (beta) for comfyui
• easy to locate: find gguf in the dropdown menu
• features: gguf loader, gguf clip loader, etc.
• others; please refer to comfyui