[v5] 🚨Refactor subprocessors handling in processors (#41633)

* remove attributes and add all missing sub processors to their auto classes

* remove all mentions of .attributes

* cleanup

* fix processor tests

* fix modular

* remove last attributes

* fixup

* fixes after merge

* fix wrong tokenizer in auto florence2

* fix missing audio_processor + nits

* Override __init__ in NewProcessor and change hf-internal-testing-repo (temporarily)

* fix auto tokenizer test

* add init to markup_lm

* update CustomProcessor in custom_processing

* remove print

* nit

* fix test modeling owlv2

* fix test_processing_layoutxlm

* Fix owlv2, wav2vec2, markuplm, voxtral issues

* add support for loading and saving multiple tokenizer natively

* remove exclude_attributes from save_pretrained

* modifs after review

Author: yonigozlan

src/transformers/models/align/processing_align.py

@@ -59,9 +59,6 @@ class AlignProcessor(ProcessorMixin):
 
     """
 
-    attributes = ["image_processor", "tokenizer"]
-    image_processor_class = "EfficientNetImageProcessor"
-    tokenizer_class = ("BertTokenizer", "BertTokenizerFast")
     valid_processor_kwargs = AlignProcessorKwargs
 
     def __init__(self, image_processor, tokenizer):

src/transformers/models/altclip/processing_altclip.py

@@ -35,10 +35,6 @@ class AltCLIPProcessor(ProcessorMixin):
             The tokenizer is a required input.
     """
 
-    attributes = ["image_processor", "tokenizer"]
-    image_processor_class = ("CLIPImageProcessor", "CLIPImageProcessorFast")
-    tokenizer_class = ("XLMRobertaTokenizer", "XLMRobertaTokenizerFast")
-
     @deprecate_kwarg(old_name="feature_extractor", version="5.0.0", new_name="image_processor")
     def __init__(self, image_processor=None, tokenizer=None):
         super().__init__(image_processor, tokenizer)

src/transformers/models/aria/modular_aria.py

@@ -906,10 +906,6 @@ class AriaProcessor(ProcessorMixin):
             A dictionary indicating size conversions for images.
     """
 
-    attributes = ["image_processor", "tokenizer"]
-    image_processor_class = "AriaImageProcessor"
-    tokenizer_class = "AutoTokenizer"
-
     def __init__(
         self,
         image_processor=None,

src/transformers/models/aria/processing_aria.py

@@ -67,10 +67,6 @@ class AriaProcessor(ProcessorMixin):
             A dictionary indicating size conversions for images.
     """
 
-    attributes = ["image_processor", "tokenizer"]
-    image_processor_class = "AriaImageProcessor"
-    tokenizer_class = "AutoTokenizer"
-
     def __init__(
         self,
         image_processor=None,

src/transformers/models/auto/configuration_auto.py

@@ -223,6 +223,7 @@
         ("layoutlm", "LayoutLMConfig"),
         ("layoutlmv2", "LayoutLMv2Config"),
         ("layoutlmv3", "LayoutLMv3Config"),
+        ("layoutxlm", "LayoutLMv2Config"),
         ("led", "LEDConfig"),
         ("levit", "LevitConfig"),
         ("lfm2", "Lfm2Config"),

src/transformers/models/auto/feature_extraction_auto.py

@@ -41,6 +41,7 @@
         ("audio-spectrogram-transformer", "ASTFeatureExtractor"),
         ("clap", "ClapFeatureExtractor"),
         ("clvp", "ClvpFeatureExtractor"),
+        ("csm", "EncodecFeatureExtractor"),
         ("dac", "DacFeatureExtractor"),
         ("data2vec-audio", "Wav2Vec2FeatureExtractor"),
         ("dia", "DiaFeatureExtractor"),
@@ -49,14 +50,20 @@
         ("granite_speech", "GraniteSpeechFeatureExtractor"),
         ("hubert", "Wav2Vec2FeatureExtractor"),
         ("kyutai_speech_to_text", "KyutaiSpeechToTextFeatureExtractor"),
+        ("markuplm", "MarkupLMFeatureExtractor"),
         ("mctct", "MCTCTFeatureExtractor"),
         ("mimi", "EncodecFeatureExtractor"),
         ("moonshine", "Wav2Vec2FeatureExtractor"),
         ("moshi", "EncodecFeatureExtractor"),
+        ("musicgen", "EncodecFeatureExtractor"),
+        ("musicgen_melody", "MusicgenMelodyFeatureExtractor"),
         ("parakeet_ctc", "ParakeetFeatureExtractor"),
         ("parakeet_encoder", "ParakeetFeatureExtractor"),
         ("phi4_multimodal", "Phi4MultimodalFeatureExtractor"),
         ("pop2piano", "Pop2PianoFeatureExtractor"),
+        ("qwen2_5_omni", "WhisperFeatureExtractor"),
+        ("qwen2_audio", "WhisperFeatureExtractor"),
+        ("qwen3_omni_moe", "WhisperFeatureExtractor"),
         ("seamless_m4t", "SeamlessM4TFeatureExtractor"),
         ("seamless_m4t_v2", "SeamlessM4TFeatureExtractor"),
         ("sew", "Wav2Vec2FeatureExtractor"),
@@ -66,6 +73,7 @@
         ("unispeech", "Wav2Vec2FeatureExtractor"),
         ("unispeech-sat", "Wav2Vec2FeatureExtractor"),
         ("univnet", "UnivNetFeatureExtractor"),
+        ("voxtral", "WhisperFeatureExtractor"),
         ("wav2vec2", "Wav2Vec2FeatureExtractor"),
         ("wav2vec2-bert", "Wav2Vec2FeatureExtractor"),
         ("wav2vec2-conformer", "Wav2Vec2FeatureExtractor"),

src/transformers/models/auto/image_processing_auto.py

@@ -62,7 +62,9 @@
             ("aimv2", ("CLIPImageProcessor", "CLIPImageProcessorFast")),
             ("aimv2_vision_model", ("CLIPImageProcessor", "CLIPImageProcessorFast")),
             ("align", ("EfficientNetImageProcessor", "EfficientNetImageProcessorFast")),
+            ("altclip", ("CLIPImageProcessor", "CLIPImageProcessorFast")),
             ("aria", ("AriaImageProcessor", None)),
+            ("aya_vision", ("GotOcr2ImageProcessor", "GotOcr2ImageProcessorFast")),
             ("beit", ("BeitImageProcessor", "BeitImageProcessorFast")),
             ("bit", ("BitImageProcessor", "BitImageProcessorFast")),
             ("blip", ("BlipImageProcessor", "BlipImageProcessorFast")),
@@ -73,6 +75,8 @@
             ("clip", ("CLIPImageProcessor", "CLIPImageProcessorFast")),
             ("clipseg", ("ViTImageProcessor", "ViTImageProcessorFast")),
             ("cohere2_vision", (None, "Cohere2VisionImageProcessorFast")),
+            ("colpali", ("SiglipImageProcessor", "SiglipImageProcessorFast")),
+            ("colqwen2", ("Qwen2VLImageProcessor", "Qwen2VLImageProcessorFast")),
             ("conditional_detr", ("ConditionalDetrImageProcessor", "ConditionalDetrImageProcessorFast")),
             ("convnext", ("ConvNextImageProcessor", "ConvNextImageProcessorFast")),
             ("convnextv2", ("ConvNextImageProcessor", "ConvNextImageProcessorFast")),
@@ -95,8 +99,10 @@
             ("efficientformer", ("EfficientFormerImageProcessor", None)),
             ("efficientloftr", ("EfficientLoFTRImageProcessor", "EfficientLoFTRImageProcessorFast")),
             ("efficientnet", ("EfficientNetImageProcessor", "EfficientNetImageProcessorFast")),
+            ("emu3", ("Emu3ImageProcessor", None)),
             ("eomt", ("EomtImageProcessor", "EomtImageProcessorFast")),
             ("flava", ("FlavaImageProcessor", "FlavaImageProcessorFast")),
+            ("florence2", ("CLIPImageProcessor", "CLIPImageProcessorFast")),
             ("focalnet", ("BitImageProcessor", "BitImageProcessorFast")),
             ("fuyu", ("FuyuImageProcessor", "FuyuImageProcessorFast")),
             ("gemma3", ("Gemma3ImageProcessor", "Gemma3ImageProcessorFast")),
@@ -114,11 +120,13 @@
             ("ijepa", ("ViTImageProcessor", "ViTImageProcessorFast")),
             ("imagegpt", ("ImageGPTImageProcessor", "ImageGPTImageProcessorFast")),
             ("instructblip", ("BlipImageProcessor", "BlipImageProcessorFast")),
+            ("internvl", ("GotOcr2ImageProcessor", "GotOcr2ImageProcessorFast")),
             ("janus", ("JanusImageProcessor", "JanusImageProcessorFast")),
             ("kosmos-2", ("CLIPImageProcessor", "CLIPImageProcessorFast")),
             ("kosmos-2.5", ("Kosmos2_5ImageProcessor", "Kosmos2_5ImageProcessorFast")),
             ("layoutlmv2", ("LayoutLMv2ImageProcessor", "LayoutLMv2ImageProcessorFast")),
             ("layoutlmv3", ("LayoutLMv3ImageProcessor", "LayoutLMv3ImageProcessorFast")),
+            ("layoutxlm", ("LayoutLMv2ImageProcessor", "LayoutLMv2ImageProcessor")),
             ("levit", ("LevitImageProcessor", "LevitImageProcessorFast")),
             ("lfm2_vl", (None, "Lfm2VlImageProcessorFast")),
             ("lightglue", ("LightGlueImageProcessor", "LightGlueImageProcessorFast")),
@@ -141,6 +149,7 @@
             ("mobilevitv2", ("MobileViTImageProcessor", "MobileViTImageProcessorFast")),
             ("nat", ("ViTImageProcessor", "ViTImageProcessorFast")),
             ("nougat", ("NougatImageProcessor", "NougatImageProcessorFast")),
+            ("omdet-turbo", ("DetrImageProcessor", "DetrImageProcessorFast")),
             ("oneformer", ("OneFormerImageProcessor", "OneFormerImageProcessorFast")),
             ("ovis2", ("Ovis2ImageProcessor", "Ovis2ImageProcessorFast")),
             ("owlv2", ("Owlv2ImageProcessor", "Owlv2ImageProcessorFast")),
@@ -155,14 +164,17 @@
             ("prompt_depth_anything", ("PromptDepthAnythingImageProcessor", "PromptDepthAnythingImageProcessorFast")),
             ("pvt", ("PvtImageProcessor", "PvtImageProcessorFast")),
             ("pvt_v2", ("PvtImageProcessor", "PvtImageProcessorFast")),
+            ("qwen2_5_omni", ("Qwen2VLImageProcessor", "Qwen2VLImageProcessorFast")),
             ("qwen2_5_vl", ("Qwen2VLImageProcessor", "Qwen2VLImageProcessorFast")),
             ("qwen2_vl", ("Qwen2VLImageProcessor", "Qwen2VLImageProcessorFast")),
+            ("qwen3_omni_moe", ("Qwen2VLImageProcessor", "Qwen2VLImageProcessorFast")),
             ("qwen3_vl", ("Qwen2VLImageProcessor", "Qwen2VLImageProcessorFast")),
             ("regnet", ("ConvNextImageProcessor", "ConvNextImageProcessorFast")),
             ("resnet", ("ConvNextImageProcessor", "ConvNextImageProcessorFast")),
             ("rt_detr", ("RTDetrImageProcessor", "RTDetrImageProcessorFast")),
             ("sam", ("SamImageProcessor", "SamImageProcessorFast")),
             ("sam2", (None, "Sam2ImageProcessorFast")),
+            ("sam2_video", (None, "Sam2ImageProcessorFast")),
             ("sam_hq", ("SamImageProcessor", "SamImageProcessorFast")),
             ("segformer", ("SegformerImageProcessor", "SegformerImageProcessorFast")),
             ("seggpt", ("SegGptImageProcessor", None)),
@@ -180,12 +192,14 @@
             ("textnet", ("TextNetImageProcessor", "TextNetImageProcessorFast")),
             ("timesformer", ("VideoMAEImageProcessor", None)),
             ("timm_wrapper", ("TimmWrapperImageProcessor", None)),
+            ("trocr", ("ViTImageProcessor", "ViTImageProcessorFast")),
             ("tvlt", ("TvltImageProcessor", None)),
             ("tvp", ("TvpImageProcessor", "TvpImageProcessorFast")),
             ("udop", ("LayoutLMv3ImageProcessor", "LayoutLMv3ImageProcessorFast")),
             ("upernet", ("SegformerImageProcessor", "SegformerImageProcessorFast")),
             ("van", ("ConvNextImageProcessor", "ConvNextImageProcessorFast")),
             ("video_llama_3", ("VideoLlama3ImageProcessor", "VideoLlama3ImageProcessorFast")),
+            ("video_llava", ("VideoLlavaImageProcessor", None)),
             ("videomae", ("VideoMAEImageProcessor", None)),
             ("vilt", ("ViltImageProcessor", "ViltImageProcessorFast")),
             ("vipllava", ("CLIPImageProcessor", "CLIPImageProcessorFast")),

src/transformers/models/auto/processing_auto.py

@@ -107,6 +107,7 @@
         ("mllama", "MllamaProcessor"),
         ("mm-grounding-dino", "GroundingDinoProcessor"),
         ("moonshine", "Wav2Vec2Processor"),
+        ("omdet-turbo", "OmDetTurboProcessor"),
         ("oneformer", "OneFormerProcessor"),
         ("ovis2", "Ovis2Processor"),
         ("owlv2", "Owlv2Processor"),