Public/automatic
1
0
Fork 0
mirror of https://github.com/vladmandic/automatic synced 2026-09-20 01:31:13 +02:00
Code Issues Packages Projects Releases Wiki Activity

SDNQ expose Dyn quant on settings

Browse Source
This commit is contained in:
Disty0
2026-01-14 16:54:50 +03:00
parent 034b0844fe
commit c2bc47e0c1
2 changed files with 8 additions and 2 deletions
Download Patch File Download Diff File Expand all files Collapse all files
modules/shared.py
+2
View File
@@ -232,7 +232,9 @@ options_templates.update(options_section(("quantization", "Model Quantization"),
"sdnq_quantize_weights_group_size": OptionInfo(0, "Group size", gr.Slider, {"minimum": -1, "maximum": 4096, "step": 1}),
"sdnq_svd_rank": OptionInfo(32, "SVD rank size", gr.Slider, {"minimum": 1, "maximum": 512, "step": 1}),
"sdnq_svd_steps": OptionInfo(8, "SVD steps", gr.Slider, {"minimum": 1, "maximum": 128, "step": 1}),
"sdnq_dynamic_loss_threshold": OptionInfo(1e-2, "Dynamic loss threshold", gr.Slider, {"minimum": 1e-4, "maximum": 1e-1, "step": 1e-4}),
"sdnq_use_svd": OptionInfo(False, "Use SVD quantization", gr.Checkbox),
"sdnq_use_dynamic_quantization": OptionInfo(False, "Use Dynamic quantization", gr.Checkbox),
"sdnq_quantize_conv_layers": OptionInfo(False, "Quantize convolutional layers", gr.Checkbox),
"sdnq_dequantize_compile": OptionInfo(devices.has_triton(early=True), "Dequantize using torch.compile", gr.Checkbox),
"sdnq_use_quantized_matmul": OptionInfo(False, "Use quantized MatMul", gr.Checkbox),