Add tooltips for DirectML & ONNX/Olive.

html/locale_en.json

@@ -244,6 +244,12 @@
   {"id":"","label":"Full precision for VAE (--no-half-vae)","localized":"","hint":"Uses FP32 for the VAE. May produce better results while using more VRAM and slower generation"},
   {"id":"","label":"Upcast sampling","localized":"","hint":"Usually produces similar results to --no-half with better performance while using less memory"},
   {"id":"","label":"Attempt VAE roll back for NaN values","localized":"","hint":"Requires Torch 2.1 and NaN check enabled"},
+  {"id":"","label":"DirectML memory stats provider","localized":"","hint":"How to get GPU memory stats"},
+  {"id":"","label":"DirectML retry ops for NaN","localized":"","hint":"Retry specific operations if their output was NaN. This may make your generation slower"},
+  {"id":"","label":"Olive use FP16 on optimization","localized":"","hint":"Use 16-bit floating point precision for the output model of Olive optimization process. Use 32-bit floating point precision if disabled"},
+  {"id":"","label":"Olive force FP32 for VAE Encoder","localized":"","hint":"Use 32-bit floating point precision for VAE Encoder of the output model. This overrides 'use FP16 on optimization' option. If you are getting NaN or black blank images from Img2Img, enable this option and remove cache"},
+  {"id":"","label":"Olive use static dimensions","localized":"","hint":"Make the inference with Olive optimized models much faster. (OrtTransformersOptimization)"},
+  {"id":"","label":"Olive cache optimized models","localized":"","hint":"Save Olive processed models as a cache. You can manage them in ONNX tab"},
   {"id":"","label":"File format","localized":"","hint":"Select file format for images"},
   {"id":"","label":"Include metadata","localized":"","hint":"Save image create parameters as metadata tags inside image file"},
   {"id":"","label":"Images filename pattern","localized":"","hint":"Use following tags to define how filenames for images are chosen:<br><pre>seq, uuid<br>date, datetime, job_timestamp<br>generation_number, batch_number<br>model, model_shortname<br>model_hash, model_name<br>sampler, seed, steps, cfg<br>clip_skip, denoising<br>hasprompt, prompt, styles<br>prompt_hash, prompt_no_styles<br>prompt_spaces, prompt_words<br>height, width, image_hash<br></pre>"},
@@ -284,6 +290,10 @@
   {"id":"","label":"VAE slicing","localized":"","hint":"Decodes batch latents one image at a time with limited VRAM. Small performance boost in VAE decode on multi-image batches"},
   {"id":"","label":"VAE tiling","localized":"","hint":"Divide large images into overlapping tiles with limited VRAM. Results in a minor increase in processing time"},
   {"id":"","label":"Attention slicing","localized":"","hint":"Performs attention computation in steps instead of all at once. Slower inference times, but greatly reduced memory usage"},
+  {"id":"","label":"Execution Provider","localized":"","hint":"ONNX Execution Provider"},
+  {"id":"","label":"ONNX show onnx-specific menu","localized":"","hint":"Enable ONNX tab in UI. Restart required"},
+  {"id":"","label":"ONNX cache converted models","localized":"","hint":"Save the models that are converted to ONNX format as a cache. You can manage them in ONNX tab"},
+  {"id":"","label":"ONNX unload base model when processing refiner","localized":"","hint":"Unload base model when the refiner is being converted/optimized/processed"},
   {"id":"","label":"inference-mode","localized":"","hint":"Use torch.inference_mode"},
   {"id":"","label":"no-grad","localized":"","hint":"Use torch.no_grad"},
   {"id":"","label":"model compile precompile","localized":"","hint":"Run model compile immediately on model load instead of first use"},

html/locale_ko.json

@@ -383,6 +383,12 @@
   {"id":"","label":"Model compile suppress errors","localized":"모델 컴파일 시 오류 억제","hint":""},
   {"id":"","label":"Disable Torch memory garbage collection","localized":"Torch 메모리 정리 비활성화","hint":"이미지 생성 시 메모리 정리를 하지 않는다. CG will still run before & after model load as well when low GPU memory threshold is reached."},
   {"id":"","label":"Enable IPEX Optimize for Intel GPUs","localized":"","hint":""},
+  {"id":"","label":"DirectML memory stats provider","localized":"DirectML GPU 메모리 정보 제공자","hint":"어떻게 GPU 메모리 정보를 가져올 것인지 정한다."},
+  {"id":"","label":"DirectML retry ops for NaN","localized":"DirectML NaN 발생 시 재시도","hint":"특정 연산에서 NaN이 발생할 시 해당 연산을 같은 파라미터로 다시 시도한다. 이미지 생성이 느려지지만 NaN이나 검은 이미지가 생성되는 것을 어느 정도 방지한다."},
+  {"id":"","label":"Olive use FP16 on optimization","localized":"Olive 최적화 시 반정밀도 사용","hint":"Olive의 OrtTransformersOptimization으로 모델을 최적화할 때 반정밀도 부동소수점(FP16)을 사용한다. 비활성화 시 단정밀도 부동소수점(FP32)을 사용한다."},
+  {"id":"","label":"Olive force FP32 for VAE Encoder","localized":"Olive VAE Encoder에 단정밀도 강제","hint":"VAE Encoder에 단정밀도 부동소수점(FP32)을 강제한다. '최적화 시 반정밀도 사용'을 무시한다. Img2Img에서 NaN이나 검은 이미지가 생성된다면 캐시를 삭제하고 VAE Encoder에 단정밀도를 사용하여 다시 시도하면 해결될 수 있다."},
+  {"id":"","label":"Olive use static dimensions","localized":"Olive 차원 고정","hint":"Olive로 최적화된 모델의 추론 속도를 매우 빠르게 만든다. (OrtTransformersOptimization)"},
+  {"id":"","label":"Olive cache optimized models","localized":"Olive 최적화된 모델 저장 (캐시)","hint":"Olive로 처리한 모델들을 캐시로 저장한다. ONNX 탭에서 관리할 수 있다."},
   {"id":"","label":"Directory for temporary images; leave empty for default","localized":"임시 이미지 저장 위치 (비워 놓으면 기본값)","hint":""},
   {"id":"","label":"Cleanup non-default temporary directory when starting webui","localized":"임시 위치가 기본값이 아닌 경우 WebUI 시작 시 비우기","hint":""},
   {"id":"","label":"Path to directory with stable diffusion checkpoints","localized":"체크포인트 위치","hint":""},
@@ -574,7 +580,11 @@
   {"id":"","label":"Enable VAE tiling","localized":"VAE 타일링 활성화","hint":"Divide large images into overlapping tiles with limited VRAM. 생성 시간이 약간 늘어날 수 있다. 어텐션 슬라이싱과 함께 사용한다."},
   {"id":"","label":"Enable attention slicing","localized":"어텐션 슬라이싱 활성화","hint":"Performs attention computation in steps instead of all at once. 생성이 10% 느려지지만 메모리 사용률을 매우 크게 줄여준다."},
   {"id":"","label":"Diffusers model loading variant","localized":"","hint":""},
-  {"id":"","label":"Diffusers VAE loading variant","localized":"","hint":""}
+  {"id":"","label":"Diffusers VAE loading variant","localized":"","hint":""},
+  {"id":"","label":"Execution Provider","localized":"","hint":"ONNX Execution Provider"},
+  {"id":"","label":"ONNX show onnx-specific menu","localized":"ONNX 탭 보이기","hint":"UI에서 ONNX 탭을 볼 수 있게 한다. WebUI를 재시작하면 반영된다."},
+  {"id":"","label":"ONNX cache converted models","localized":"ONNX 모델 저장 (캐시)","hint":"ONNX 형식으로 변환된 모델을 캐시로 저장한다. ONNX 탭에서 관리할 수 있다."},
+  {"id":"","label":"ONNX unload base model when processing refiner","localized":"ONNX 리파이너를 처리하고 있을 때 base 모델을 메모리에서 해제","hint":"리파이너 모델이 변환, 최적화 또는 처리되고 있을 때 base 모델을 메모리에서 잠시 해제합니다."}
 ],
 "scripts": [
   {"id":"","label":"Script","localized":"스크립트","hint":""},

modules/shared.py

@@ -389,9 +389,9 @@ options_templates.update(options_section(('cuda', "Compute Settings"), {
     "directml_catch_nan": OptionInfo(False, "DirectML retry ops for NaN", gr.Checkbox, {"visible": devices.backend == "directml"}),
 
     "olive_sep": OptionInfo("<h2>Olive</h2>", "", gr.HTML),
-    "olive_float16": OptionInfo(True, 'Olive use FP16 on optimization (will use FP32 if unchecked)'),
-    "olive_vae_encoder_float32": OptionInfo(False, 'Olive force FP32 for VAE Encoder (if Img2Img generates NaN, enable this option and remove previously optimized model)'),
-    "olive_static_dims": OptionInfo(True, 'Olive use static dimensions (make inference faster with OrtTransformersOptimization)'),
+    "olive_float16": OptionInfo(True, 'Olive use FP16 on optimization'),
+    "olive_vae_encoder_float32": OptionInfo(False, 'Olive force FP32 for VAE Encoder'),
+    "olive_static_dims": OptionInfo(True, 'Olive use static dimensions'),
     "olive_cache_optimized": OptionInfo(True, 'Olive cache optimized models'),
 }))
 
@@ -446,7 +446,7 @@ options_templates.update(options_section(('diffusers', "Diffusers Settings"), {
 
     "onnx_sep": OptionInfo("<h2>ONNX Runtime</h2>", "", gr.HTML),
     "onnx_execution_provider": OptionInfo(get_default_execution_provider().value, 'Execution Provider', gr.Dropdown, lambda: {"choices": available_execution_providers }),
-    "onnx_show_menu": OptionInfo(False, 'ONNX show onnx-specific menu (restart required)'),
+    "onnx_show_menu": OptionInfo(False, 'ONNX show onnx-specific menu'),
     "onnx_cache_converted": OptionInfo(True, 'ONNX cache converted models'),
     "onnx_unload_base": OptionInfo(False, 'ONNX unload base model when processing refiner'),
 }))