diff --git a/modules/shared.py b/modules/shared.py index 91def4d6e..a79382ca2 100644 --- a/modules/shared.py +++ b/modules/shared.py @@ -522,11 +522,6 @@ options_templates.update(options_section(('backends', "Backend Settings"), { })) options_templates.update(options_section(('quantization', "Quantization Settings"), { - "bnb_quantization_sep": OptionInfo("

BitsAndBytes

", "", gr.HTML), - "bnb_quantization": OptionInfo([], "Quantization enabled", gr.CheckboxGroup, {"choices": ["Model", "Transformer", "VAE", "TE", "Video", "LLM", "ControlNet"], "visible": native}), - "bnb_quantization_type": OptionInfo("nf4", "Quantization type", gr.Dropdown, {"choices": ['nf4', 'fp8', 'fp4'], "visible": native}), - "bnb_quantization_storage": OptionInfo("uint8", "Backend storage", gr.Dropdown, {"choices": ["float16", "float32", "int8", "uint8", "float64", "bfloat16"], "visible": native}), - "sdnq_quantize_sep": OptionInfo("

SDNQ: SDNext Quantization

", "", gr.HTML), "sdnq_quantize_weights": OptionInfo([], "Quantization enabled", gr.CheckboxGroup, {"choices": ["Model", "Transformer", "VAE", "TE", "Video", "LLM", "ControlNet"], "visible": native}), "sdnq_quantize_mode": OptionInfo("pre", "Quantization mode", gr.Dropdown, {"choices": ['pre', 'post'], "visible": native}), @@ -538,6 +533,11 @@ options_templates.update(options_section(('quantization', "Quantization Settings "sdnq_decompress_int8_matmul": OptionInfo(False, "Use direct INT8 MatMul", gr.Checkbox, {"visible": native}), "sdnq_quantize_shuffle_weights": OptionInfo(False, "Shuffle weights in post mode", gr.Checkbox, {"visible": native}), + "bnb_quantization_sep": OptionInfo("

BitsAndBytes

", "", gr.HTML), + "bnb_quantization": OptionInfo([], "Quantization enabled", gr.CheckboxGroup, {"choices": ["Model", "Transformer", "VAE", "TE", "Video", "LLM", "ControlNet"], "visible": native}), + "bnb_quantization_type": OptionInfo("nf4", "Quantization type", gr.Dropdown, {"choices": ['nf4', 'fp8', 'fp4'], "visible": native}), + "bnb_quantization_storage": OptionInfo("uint8", "Backend storage", gr.Dropdown, {"choices": ["float16", "float32", "int8", "uint8", "float64", "bfloat16"], "visible": native}), + "quanto_quantization_sep": OptionInfo("

Optimum Quanto

", "", gr.HTML), "quanto_quantization": OptionInfo([], "Quantization enabled", gr.CheckboxGroup, {"choices": ["Model", "Transformer", "VAE", "TE", "Video", "LLM", "ControlNet"], "visible": native}), "quanto_quantization_type": OptionInfo("int8", "Quantization weights type", gr.Dropdown, {"choices": ["float8", "int8", "int4", "int2"], "visible": native}),