OpenVINO Quantization support with NNCF

This commit is contained in:
Disty0
2024-01-25 20:22:57 +03:00
parent d1bb51eee0
commit db6fd95351
4 changed files with 78 additions and 44 deletions
+3 -1
View File
@@ -378,7 +378,9 @@ options_templates.update(options_section(('cuda', "Compute Settings"), {
"openvino_sep": OptionInfo("<h2>OpenVINO</h2>", "", gr.HTML, {"visible": cmd_opts.use_openvino}),
"openvino_devices": OptionInfo([], "OpenVINO devices to use", gr.CheckboxGroup, {"choices": get_openvino_device_list() if cmd_opts.use_openvino else [], "visible": cmd_opts.use_openvino}),
"nncf_compress_weights_mode": OptionInfo("INT8", "OpenVINO compress mode for NNCF", gr.Radio, {"choices": ['INT8', 'INT4_SYM', 'INT4_ASYM', 'NF4'], "visible": cmd_opts.use_openvino}),
"nncf_quantize": OptionInfo([], "OpenVINO Quantize Models with NNCF", gr.CheckboxGroup, {"choices": ["Model", "VAE", "Text Encoder"], "visible": cmd_opts.use_openvino}),
"nncf_quant_mode": OptionInfo("INT8", "OpenVINO Quantization Mode with NNCF", gr.Radio, {"choices": ['INT8', 'FP8_E4M3', 'FP8_E5M2'], "visible": cmd_opts.use_openvino}),
"nncf_compress_weights_mode": OptionInfo("INT8", "OpenVINO compress mode for NNCF", gr.Radio, {"choices": ['INT8', 'INT8_SYM', 'INT4_ASYM', 'INT4_SYM', 'NF4'], "visible": cmd_opts.use_openvino}),
"nncf_compress_weights_raito": OptionInfo(1.0, "OpenVINO compress ratio for NNCF", gr.Slider, {"minimum": 0, "maximum": 1, "step": 0.01, "visible": cmd_opts.use_openvino}),
"openvino_disable_model_caching": OptionInfo(False, "OpenVINO disable model caching", gr.Checkbox, {"visible": cmd_opts.use_openvino}),
}))