从预训练模型配置实例化库的配置类之一。

要实例化的配置类是根据加载的配置对象的 model_type 属性选择的，或者在缺少该属性时，通过回退到在 pretrained_model_name_or_path 上使用模式匹配来选择。

• albert — AlbertConfig (ALBERT 模型)
• align — AlignConfig (ALIGN 模型)
• altclip — AltCLIPConfig (AltCLIP 模型)
• aria — AriaConfig (Aria 模型)
• aria_text — AriaTextConfig (AriaText 模型)
• audio-spectrogram-transformer — ASTConfig (Audio Spectrogram Transformer 模型)
• autoformer — AutoformerConfig (Autoformer 模型)
• aya_vision — AyaVisionConfig (AyaVision 模型)
• bamba — BambaConfig (Bamba 模型)
• bark — BarkConfig (Bark 模型)
• bart — BartConfig (BART 模型)
• beit — BeitConfig (BEiT 模型)
• bert — BertConfig (BERT 模型)
• bert-generation — BertGenerationConfig (Bert Generation 模型)
• big_bird — BigBirdConfig (BigBird 模型)
• bigbird_pegasus — BigBirdPegasusConfig (BigBird-Pegasus 模型)
• biogpt — BioGptConfig (BioGpt 模型)
• bit — BitConfig (BiT 模型)
• blenderbot — BlenderbotConfig (Blenderbot 模型)
• blenderbot-small — BlenderbotSmallConfig (BlenderbotSmall 模型)
• blip — BlipConfig (BLIP 模型)
• blip-2 — Blip2Config (BLIP-2 模型)
• bloom — BloomConfig (BLOOM 模型)
• bridgetower — BridgeTowerConfig (BridgeTower 模型)
• bros — BrosConfig (BROS 模型)
• camembert — CamembertConfig (CamemBERT 模型)
• canine — CanineConfig (CANINE 模型)
• chameleon — ChameleonConfig (Chameleon 模型)
• chinese_clip — ChineseCLIPConfig (Chinese-CLIP 模型)
• chinese_clip_vision_model — ChineseCLIPVisionConfig (ChineseCLIPVisionModel 模型)
• clap — ClapConfig (CLAP 模型)
• clip — CLIPConfig (CLIP 模型)
• clip_text_model — CLIPTextConfig (CLIPTextModel 模型)
• clip_vision_model — CLIPVisionConfig (CLIPVisionModel 模型)
• clipseg — CLIPSegConfig (CLIPSeg 模型)
• clvp — ClvpConfig (CLVP 模型)
• code_llama — LlamaConfig (CodeLlama 模型)
• codegen — CodeGenConfig (CodeGen 模型)
• cohere — CohereConfig (Cohere 模型)
• cohere2 — Cohere2Config (Cohere2 模型)
• colpali — ColPaliConfig (ColPali 模型)
• conditional_detr — ConditionalDetrConfig (Conditional DETR 模型)
• convbert — ConvBertConfig (ConvBERT 模型)
• convnext — ConvNextConfig (ConvNeXT 模型)
• convnextv2 — ConvNextV2Config (ConvNeXTV2 模型)
• cpmant — CpmAntConfig (CPM-Ant 模型)
• ctrl — CTRLConfig (CTRL 模型)
• cvt — CvtConfig (CvT 模型)
• dab-detr — DabDetrConfig (DAB-DETR 模型)
• dac — DacConfig (DAC 模型)
• data2vec-audio — Data2VecAudioConfig (Data2VecAudio 模型)
• data2vec-text — Data2VecTextConfig (Data2VecText 模型)
• data2vec-vision — Data2VecVisionConfig (Data2VecVision 模型)
• dbrx — DbrxConfig (DBRX 模型)
• deberta — DebertaConfig (DeBERTa 模型)
• deberta-v2 — DebertaV2Config (DeBERTa-v2 模型)
• decision_transformer — DecisionTransformerConfig (Decision Transformer 模型)
• deformable_detr — DeformableDetrConfig (Deformable DETR 模型)
• deit — DeiTConfig (DeiT 模型)
• depth_anything — DepthAnythingConfig (Depth Anything 模型)
• depth_pro — DepthProConfig (DepthPro 模型)
• deta — DetaConfig (DETA 模型)
• detr — DetrConfig (DETR 模型)
• diffllama — DiffLlamaConfig (DiffLlama 模型)
• dinat — DinatConfig (DiNAT 模型)
• dinov2 — Dinov2Config (DINOv2 模型)
• dinov2_with_registers — Dinov2WithRegistersConfig (DINOv2 with Registers 模型)
• distilbert — DistilBertConfig (DistilBERT 模型)
• donut-swin — DonutSwinConfig (DonutSwin 模型)
• dpr — DPRConfig (DPR 模型)
• dpt — DPTConfig (DPT 模型)
• efficientformer — EfficientFormerConfig (EfficientFormer 模型)
• efficientnet — EfficientNetConfig (EfficientNet 模型)
• electra — ElectraConfig (ELECTRA 模型)
• emu3 — Emu3Config (Emu3 模型)
• encodec — EncodecConfig (EnCodec 模型)
• encoder-decoder — EncoderDecoderConfig (Encoder decoder 模型)
• ernie — ErnieConfig (ERNIE 模型)
• ernie_m — ErnieMConfig (ErnieM 模型)
• esm — EsmConfig (ESM 模型)
• falcon — FalconConfig (Falcon 模型)
• falcon_mamba — FalconMambaConfig (FalconMamba 模型)
• fastspeech2_conformer — FastSpeech2ConformerConfig (FastSpeech2Conformer 模型)
• flaubert — FlaubertConfig (FlauBERT 模型)
• flava — FlavaConfig (FLAVA 模型)
• fnet — FNetConfig (FNet 模型)
• focalnet — FocalNetConfig (FocalNet 模型)
• fsmt — FSMTConfig (FairSeq 机器翻译模型)
• funnel — FunnelConfig (Funnel Transformer 模型)
• fuyu — FuyuConfig (Fuyu 模型)
• gemma — GemmaConfig (Gemma 模型)
• gemma2 — Gemma2Config (Gemma2 模型)
• gemma3 — Gemma3Config (Gemma3ForConditionalGeneration 模型)
• gemma3_text — Gemma3TextConfig (Gemma3ForCausalLM 模型)
• git — GitConfig (GIT 模型)
• glm — GlmConfig (GLM 模型)
• glpn — GLPNConfig (GLPN 模型)
• got_ocr2 — GotOcr2Config (GOT-OCR2 模型)
• gpt-sw3 — GPT2Config (GPT-Sw3 模型)
• gpt2 — GPT2Config (OpenAI GPT-2 模型)
• gpt_bigcode — GPTBigCodeConfig (GPTBigCode 模型)
• gpt_neo — GPTNeoConfig (GPT Neo 模型)
• gpt_neox — GPTNeoXConfig (GPT NeoX 模型)
• gpt_neox_japanese — GPTNeoXJapaneseConfig (GPT NeoX Japanese 模型)
• gptj — GPTJConfig (GPT-J 模型)
• gptsan-japanese — GPTSanJapaneseConfig (GPTSAN-japanese 模型)
• granite — GraniteConfig (Granite 模型)
• granitemoe — GraniteMoeConfig (GraniteMoeMoe 模型)
• granitemoeshared — GraniteMoeSharedConfig (GraniteMoeSharedMoe 模型)
• granitevision — LlavaNextConfig (LLaVA-NeXT 模型)
• graphormer — GraphormerConfig (Graphormer 模型)
• grounding-dino — GroundingDinoConfig (Grounding DINO 模型)
• groupvit — GroupViTConfig (GroupViT 模型)
• helium — HeliumConfig (Helium 模型)
• hiera — HieraConfig (Hiera 模型)
• hubert — HubertConfig (Hubert 模型)
• ibert — IBertConfig (I-BERT 模型)
• idefics — IdeficsConfig (IDEFICS 模型)
• idefics2 — Idefics2Config (Idefics2 模型)
• idefics3 — Idefics3Config (Idefics3 模型)
• idefics3_vision — Idefics3VisionConfig (Idefics3VisionTransformer 模型)
• ijepa — IJepaConfig (I-JEPA 模型)
• imagegpt — ImageGPTConfig (ImageGPT 模型)
• informer — InformerConfig (Informer 模型)
• instructblip — InstructBlipConfig (InstructBLIP 模型)
• instructblipvideo — InstructBlipVideoConfig (InstructBlipVideo 模型)
• jamba — JambaConfig (Jamba 模型)
• jetmoe — JetMoeConfig (JetMoe 模型)
• jukebox — JukeboxConfig (Jukebox 模型)
• kosmos-2 — Kosmos2Config (KOSMOS-2 模型)
• layoutlm — LayoutLMConfig (LayoutLM 模型)
• layoutlmv2 — LayoutLMv2Config (LayoutLMv2 模型)
• layoutlmv3 — LayoutLMv3Config (LayoutLMv3 模型)
• led — LEDConfig (LED 模型)
• levit — LevitConfig (LeViT 模型)
• lilt — LiltConfig (LiLT 模型)
• llama — LlamaConfig (LLaMA 模型)
• llava — LlavaConfig (LLaVa 模型)
• llava_next — LlavaNextConfig (LLaVA-NeXT 模型)
• llava_next_video — LlavaNextVideoConfig (LLaVa-NeXT-Video 模型)
• llava_onevision — LlavaOnevisionConfig (LLaVA-Onevision 模型)
• longformer — LongformerConfig (Longformer 模型)
• longt5 — LongT5Config (LongT5 模型)
• luke — LukeConfig (LUKE 模型)
• lxmert — LxmertConfig (LXMERT 模型)
• m2m_100 — M2M100Config (M2M100 模型)
• mamba — MambaConfig (Mamba 模型)
• mamba2 — Mamba2Config (mamba2 模型)
• marian — MarianConfig (Marian 模型)
• markuplm — MarkupLMConfig (MarkupLM 模型)
• mask2former — Mask2FormerConfig (Mask2Former 模型)
• maskformer — MaskFormerConfig (MaskFormer 模型)
• maskformer-swin — MaskFormerSwinConfig (MaskFormerSwin 模型)
• mbart — MBartConfig (mBART 模型)
• mctct — MCTCTConfig (M-CTC-T 模型)
• mega — MegaConfig (MEGA 模型)
• megatron-bert — MegatronBertConfig (Megatron-BERT 模型)
• mgp-str — MgpstrConfig (MGP-STR 模型)
• mimi — MimiConfig (Mimi 模型)
• mistral — MistralConfig (Mistral 模型)
• mistral3 — Mistral3Config (Mistral3 模型)
• mixtral — MixtralConfig (Mixtral 模型)
• mllama — MllamaConfig (Mllama 模型)
• mobilebert — MobileBertConfig (MobileBERT 模型)
• mobilenet_v1 — MobileNetV1Config (MobileNetV1 模型)
• mobilenet_v2 — MobileNetV2Config (MobileNetV2 模型)
• mobilevit — MobileViTConfig (MobileViT 模型)
• mobilevitv2 — MobileViTV2Config (MobileViTV2 模型)
• modernbert — ModernBertConfig (ModernBERT 模型)
• moonshine — MoonshineConfig (Moonshine 模型)
• moshi — MoshiConfig (Moshi 模型)
• mpnet — MPNetConfig (MPNet 模型)
• mpt — MptConfig (MPT 模型)
• mra — MraConfig (MRA 模型)
• mt5 — MT5Config (MT5 模型)
• musicgen — MusicgenConfig (MusicGen 模型)
• musicgen_melody — MusicgenMelodyConfig (MusicGen Melody 模型)
• mvp — MvpConfig (MVP 模型)
• nat — NatConfig (NAT 模型)
• nemotron — NemotronConfig (Nemotron 模型)
• nezha — NezhaConfig (Nezha 模型)
• nllb-moe — NllbMoeConfig (NLLB-MOE 模型)
• nougat — VisionEncoderDecoderConfig (Nougat 模型)
• nystromformer — NystromformerConfig (Nyströmformer 模型)
• olmo — OlmoConfig (OLMo 模型)
• olmo2 — Olmo2Config (OLMo2 模型)
• olmoe — OlmoeConfig (OLMoE 模型)
• omdet-turbo — OmDetTurboConfig (OmDet-Turbo 模型)
• oneformer — OneFormerConfig (OneFormer 模型)
• open-llama — OpenLlamaConfig (OpenLlama 模型)
• openai-gpt — OpenAIGPTConfig (OpenAI GPT 模型)
• opt — OPTConfig (OPT 模型)
• owlv2 — Owlv2Config (OWLv2 模型)
• owlvit — OwlViTConfig (OWL-ViT 模型)
• paligemma — PaliGemmaConfig (PaliGemma 模型)
• patchtsmixer — PatchTSMixerConfig (PatchTSMixer 模型)
• patchtst — PatchTSTConfig (PatchTST 模型)
• pegasus — PegasusConfig (Pegasus 模型)
• pegasus_x — PegasusXConfig (PEGASUS-X 模型)
• perceiver — PerceiverConfig (Perceiver 模型)
• persimmon — PersimmonConfig (Persimmon 模型)
• phi — PhiConfig (Phi 模型)
• phi3 — Phi3Config (Phi3 模型)
• phimoe — PhimoeConfig (Phimoe 模型)
• pix2struct — Pix2StructConfig (Pix2Struct 模型)
• pixtral — PixtralVisionConfig (Pixtral 模型)
• plbart — PLBartConfig (PLBart 模型)
• poolformer — PoolFormerConfig (PoolFormer 模型)
• pop2piano — Pop2PianoConfig (Pop2Piano 模型)
• prompt_depth_anything — PromptDepthAnythingConfig (PromptDepthAnything 模型)
• prophetnet — ProphetNetConfig (ProphetNet 模型)
• pvt — PvtConfig (PVT 模型)
• pvt_v2 — PvtV2Config (PVTv2 模型)
• qdqbert — QDQBertConfig (QDQBert 模型)
• qwen2 — Qwen2Config (Qwen2 模型)
• qwen2_5_vl — Qwen2_5_VLConfig (Qwen2_5_VL 模型)
• qwen2_audio — Qwen2AudioConfig (Qwen2Audio 模型)
• qwen2_audio_encoder — Qwen2AudioEncoderConfig (Qwen2AudioEncoder 模型)
• qwen2_moe — Qwen2MoeConfig (Qwen2MoE 模型)
• qwen2_vl — Qwen2VLConfig (Qwen2VL 模型)
• rag — RagConfig (RAG 模型)
• realm — RealmConfig (REALM 模型)
• recurrent_gemma — RecurrentGemmaConfig (RecurrentGemma 模型)
• reformer — ReformerConfig (Reformer 模型)
• regnet — RegNetConfig (RegNet 模型)
• rembert — RemBertConfig (RemBERT 模型)
• resnet — ResNetConfig (ResNet 模型)
• retribert — RetriBertConfig (RetriBERT 模型)
• roberta — RobertaConfig (RoBERTa 模型)
• roberta-prelayernorm — RobertaPreLayerNormConfig (RoBERTa-PreLayerNorm 模型)
• roc_bert — RoCBertConfig (RoCBert 模型)
• roformer — RoFormerConfig (RoFormer 模型)
• rt_detr — RTDetrConfig (RT-DETR 模型)
• rt_detr_resnet — RTDetrResNetConfig (RT-DETR-ResNet 模型)
• rt_detr_v2 — RTDetrV2Config (RT-DETRv2 模型)
• rwkv — RwkvConfig (RWKV 模型)
• sam — SamConfig (SAM 模型)
• seamless_m4t — SeamlessM4TConfig (SeamlessM4T 模型)
• seamless_m4t_v2 — SeamlessM4Tv2Config (SeamlessM4Tv2 模型)
• segformer — SegformerConfig (SegFormer 模型)
• seggpt — SegGptConfig (SegGPT 模型)
• sew — SEWConfig (SEW 模型)
• sew-d — SEWDConfig (SEW-D 模型)
• shieldgemma2 — ShieldGemma2Config (Shieldgemma2 模型)
• siglip — SiglipConfig (SigLIP 模型)
• siglip2 — Siglip2Config (SigLIP2 模型)
• siglip_vision_model — SiglipVisionConfig (SiglipVisionModel 模型)
• smolvlm — SmolVLMConfig (SmolVLM 模型)
• smolvlm_vision — SmolVLMVisionConfig (SmolVLMVisionTransformer 模型)
• speech-encoder-decoder — SpeechEncoderDecoderConfig (Speech Encoder decoder 模型)
• speech_to_text — Speech2TextConfig (Speech2Text 模型)
• speech_to_text_2 — Speech2Text2Config (Speech2Text2 模型)
• speecht5 — SpeechT5Config (SpeechT5 模型)
• splinter — SplinterConfig (Splinter 模型)
• squeezebert — SqueezeBertConfig (SqueezeBERT 模型)
• stablelm — StableLmConfig (StableLm 模型)
• starcoder2 — Starcoder2Config (Starcoder2 模型)
• superglue — SuperGlueConfig (SuperGlue 模型)
• superpoint — SuperPointConfig (SuperPoint 模型)
• swiftformer — SwiftFormerConfig (SwiftFormer 模型)
• swin — SwinConfig (Swin Transformer 模型)
• swin2sr — Swin2SRConfig (Swin2SR 模型)
• swinv2 — Swinv2Config (Swin Transformer V2 模型)
• switch_transformers — SwitchTransformersConfig (SwitchTransformers 模型)
• t5 — T5Config (T5 模型)
• table-transformer — TableTransformerConfig (Table Transformer 模型)
• tapas — TapasConfig (TAPAS 模型)
• textnet — TextNetConfig (TextNet 模型)
• time_series_transformer — TimeSeriesTransformerConfig (Time Series Transformer 模型)
• timesformer — TimesformerConfig (TimeSformer 模型)
• timm_backbone — TimmBackboneConfig (TimmBackbone 模型)
• timm_wrapper — TimmWrapperConfig (TimmWrapperModel 模型)
• trajectory_transformer — TrajectoryTransformerConfig (Trajectory Transformer 模型)
• transfo-xl — TransfoXLConfig (Transformer-XL 模型)
• trocr — TrOCRConfig (TrOCR 模型)
• tvlt — TvltConfig (TVLT 模型)
• tvp — TvpConfig (TVP 模型)
• udop — UdopConfig (UDOP 模型)
• umt5 — UMT5Config (UMT5 模型)
• unispeech — UniSpeechConfig (UniSpeech 模型)
• unispeech-sat — UniSpeechSatConfig (UniSpeechSat 模型)
• univnet — UnivNetConfig (UnivNet 模型)
• upernet — UperNetConfig (UPerNet 模型)
• van — VanConfig (VAN 模型)
• video_llava — VideoLlavaConfig (VideoLlava 模型)
• videomae — VideoMAEConfig (VideoMAE 模型)
• vilt — ViltConfig (ViLT 模型)
• vipllava — VipLlavaConfig (VipLlava 模型)
• vision-encoder-decoder — VisionEncoderDecoderConfig (Vision Encoder decoder 模型)
• vision-text-dual-encoder — VisionTextDualEncoderConfig (VisionTextDualEncoder 模型)
• visual_bert — VisualBertConfig (VisualBERT 模型)
• vit — ViTConfig (ViT 模型)
• vit_hybrid — ViTHybridConfig (ViT Hybrid 模型)
• vit_mae — ViTMAEConfig (ViTMAE 模型)
• vit_msn — ViTMSNConfig (ViTMSN 模型)
• vitdet — VitDetConfig (VitDet 模型)
• vitmatte — VitMatteConfig (ViTMatte 模型)
• vitpose — VitPoseConfig (ViTPose 模型)
• vitpose_backbone — VitPoseBackboneConfig (ViTPoseBackbone 模型)
• vits — VitsConfig (VITS 模型)
• vivit — VivitConfig (ViViT 模型)
• wav2vec2 — Wav2Vec2Config (Wav2Vec2 模型)
• wav2vec2-bert — Wav2Vec2BertConfig (Wav2Vec2-BERT 模型)
• wav2vec2-conformer — Wav2Vec2ConformerConfig (Wav2Vec2-Conformer 模型)
• wavlm — WavLMConfig (WavLM 模型)
• whisper — WhisperConfig (Whisper 模型)
• xclip — XCLIPConfig (X-CLIP 模型)
• xglm — XGLMConfig (XGLM 模型)
• xlm — XLMConfig (XLM 模型)
• xlm-prophetnet — XLMProphetNetConfig (XLM-ProphetNet 模型)
• xlm-roberta — XLMRobertaConfig (XLM-RoBERTa 模型)
• xlm-roberta-xl — XLMRobertaXLConfig (XLM-RoBERTa-XL 模型)
• xlnet — XLNetConfig (XLNet 模型)
• xmod — XmodConfig (X-MOD 模型)
• yolos — YolosConfig (YOLOS 模型)
• yoso — YosoConfig (YOSO 模型)
• zamba — ZambaConfig (Zamba 模型)
• zamba2 — Zamba2Config (Zamba2 模型)
• zoedepth — ZoeDepthConfig (ZoeDepth 模型)

>>> from transformers import AutoConfig

>>> # Download configuration from huggingface.co and cache.
>>> config = AutoConfig.from_pretrained("google-bert/bert-base-uncased")

>>> # Download configuration from huggingface.co (user-uploaded) and cache.
>>> config = AutoConfig.from_pretrained("dbmdz/bert-base-german-cased")

>>> # If configuration file is in a directory (e.g., was saved using *save_pretrained('./test/saved_model/')*).
>>> config = AutoConfig.from_pretrained("./test/bert_saved_model/")

>>> # Load a specific configuration file.
>>> config = AutoConfig.from_pretrained("./test/bert_saved_model/my_configuration.json")

>>> # Change some config attributes when loading a pretrained config.
>>> config = AutoConfig.from_pretrained("google-bert/bert-base-uncased", output_attentions=True, foo=False)
>>> config.output_attentions
True

>>> config, unused_kwargs = AutoConfig.from_pretrained(
...     "google-bert/bert-base-uncased", output_attentions=True, foo=False, return_unused_kwargs=True
... )
>>> config.output_attentions
True

>>> unused_kwargs
{'foo': False}

为此类注册新的配置。

从预训练模型词汇表实例化库中的一个分词器类。

要实例化的分词器类是基于配置对象的 model_type 属性选择的（可以作为参数传递，也可以在可能的情况下从 pretrained_model_name_or_path 加载），或者在缺少该属性时，通过回退到对 pretrained_model_name_or_path 使用模式匹配来选择。

• albert — AlbertTokenizer 或 AlbertTokenizerFast (ALBERT 模型)
• align — BertTokenizer 或 BertTokenizerFast (ALIGN 模型)
• aria — LlamaTokenizer 或 LlamaTokenizerFast (Aria 模型)
• aya_vision — CohereTokenizerFast (AyaVision 模型)
• bark — BertTokenizer 或 BertTokenizerFast (Bark 模型)
• bart — BartTokenizer 或 BartTokenizerFast (BART 模型)
• barthez — BarthezTokenizer 或 BarthezTokenizerFast (BARThez 模型)
• bartpho — BartphoTokenizer (BARTpho 模型)
• bert — BertTokenizer 或 BertTokenizerFast (BERT 模型)
• bert-generation — BertGenerationTokenizer (Bert Generation 模型)
• bert-japanese — BertJapaneseTokenizer (BertJapanese 模型)
• bertweet — BertweetTokenizer (BERTweet 模型)
• big_bird — BigBirdTokenizer 或 BigBirdTokenizerFast (BigBird 模型)
• bigbird_pegasus — PegasusTokenizer 或 PegasusTokenizerFast (BigBird-Pegasus 模型)
• biogpt — BioGptTokenizer (BioGpt 模型)
• blenderbot — BlenderbotTokenizer 或 BlenderbotTokenizerFast (Blenderbot 模型)
• blenderbot-small — BlenderbotSmallTokenizer (BlenderbotSmall 模型)
• blip — BertTokenizer 或 BertTokenizerFast (BLIP 模型)
• blip-2 — GPT2Tokenizer 或 GPT2TokenizerFast (BLIP-2 模型)
• bloom — BloomTokenizerFast (BLOOM 模型)
• bridgetower — RobertaTokenizer 或 RobertaTokenizerFast (BridgeTower 模型)
• bros — BertTokenizer 或 BertTokenizerFast (BROS 模型)
• byt5 — ByT5Tokenizer (ByT5 模型)
• camembert — CamembertTokenizer 或 CamembertTokenizerFast (CamemBERT 模型)
• canine — CanineTokenizer (CANINE 模型)
• chameleon — LlamaTokenizer 或 LlamaTokenizerFast (Chameleon 模型)
• chinese_clip — BertTokenizer 或 BertTokenizerFast (Chinese-CLIP 模型)
• clap — RobertaTokenizer 或 RobertaTokenizerFast (CLAP 模型)
• clip — CLIPTokenizer 或 CLIPTokenizerFast (CLIP 模型)
• clipseg — CLIPTokenizer 或 CLIPTokenizerFast (CLIPSeg 模型)
• clvp — ClvpTokenizer (CLVP 模型)
• code_llama — CodeLlamaTokenizer 或 CodeLlamaTokenizerFast (CodeLlama 模型)
• codegen — CodeGenTokenizer 或 CodeGenTokenizerFast (CodeGen 模型)
• cohere — CohereTokenizerFast (Cohere 模型)
• cohere2 — CohereTokenizerFast (Cohere2 模型)
• colpali — LlamaTokenizer 或 LlamaTokenizerFast (ColPali 模型)
• convbert — ConvBertTokenizer 或 ConvBertTokenizerFast (ConvBERT 模型)
• cpm — CpmTokenizer 或 CpmTokenizerFast (CPM 模型)
• cpmant — CpmAntTokenizer (CPM-Ant 模型)
• ctrl — CTRLTokenizer (CTRL 模型)
• data2vec-audio — Wav2Vec2CTCTokenizer (Data2VecAudio 模型)
• data2vec-text — RobertaTokenizer 或 RobertaTokenizerFast (Data2VecText 模型)
• dbrx — GPT2Tokenizer 或 GPT2TokenizerFast (DBRX 模型)
• deberta — DebertaTokenizer 或 DebertaTokenizerFast (DeBERTa 模型)
• deberta-v2 — DebertaV2Tokenizer 或 DebertaV2TokenizerFast (DeBERTa-v2 模型)
• diffllama — LlamaTokenizer 或 LlamaTokenizerFast (DiffLlama 模型)
• distilbert — DistilBertTokenizer 或 DistilBertTokenizerFast (DistilBERT 模型)
• dpr — DPRQuestionEncoderTokenizer 或 DPRQuestionEncoderTokenizerFast (DPR 模型)
• electra — ElectraTokenizer 或 ElectraTokenizerFast (ELECTRA 模型)
• emu3 — GPT2Tokenizer 或 GPT2TokenizerFast (Emu3 模型)
• ernie — BertTokenizer 或 BertTokenizerFast (ERNIE 模型)
• ernie_m — ErnieMTokenizer (ErnieM 模型)
• esm — EsmTokenizer (ESM 模型)
• falcon — PreTrainedTokenizerFast (Falcon 模型)
• falcon_mamba — GPTNeoXTokenizerFast (FalconMamba 模型)
• fastspeech2_conformer — (FastSpeech2Conformer 模型)
• flaubert — FlaubertTokenizer (FlauBERT 模型)
• fnet — FNetTokenizer 或 FNetTokenizerFast (FNet 模型)
• fsmt — FSMTTokenizer (FairSeq 机器翻译模型)
• funnel — FunnelTokenizer 或 FunnelTokenizerFast (Funnel Transformer 模型)
• gemma — GemmaTokenizer 或 GemmaTokenizerFast (Gemma 模型)
• gemma2 — GemmaTokenizer 或 GemmaTokenizerFast (Gemma2 模型)
• gemma3 — GemmaTokenizer 或 GemmaTokenizerFast (Gemma3ForConditionalGeneration 模型)
• gemma3_text — GemmaTokenizer 或 GemmaTokenizerFast (Gemma3ForCausalLM 模型)
• git — BertTokenizer 或 BertTokenizerFast (GIT 模型)
• glm — PreTrainedTokenizerFast (GLM 模型)
• gpt-sw3 — GPTSw3Tokenizer (GPT-Sw3 模型)
• gpt2 — GPT2Tokenizer 或 GPT2TokenizerFast (OpenAI GPT-2 模型)
• gpt_bigcode — GPT2Tokenizer 或 GPT2TokenizerFast (GPTBigCode 模型)
• gpt_neo — GPT2Tokenizer 或 GPT2TokenizerFast (GPT Neo 模型)
• gpt_neox — GPTNeoXTokenizerFast (GPT NeoX 模型)
• gpt_neox_japanese — GPTNeoXJapaneseTokenizer (GPT NeoX 日语模型)
• gptj — GPT2Tokenizer 或 GPT2TokenizerFast (GPT-J 模型)
• gptsan-japanese — GPTSanJapaneseTokenizer (GPTSAN-日语模型)
• grounding-dino — BertTokenizer 或 BertTokenizerFast (Grounding DINO 模型)
• groupvit — CLIPTokenizer 或 CLIPTokenizerFast (GroupViT 模型)
• helium — PreTrainedTokenizerFast (Helium 模型)
• herbert — HerbertTokenizer 或 HerbertTokenizerFast (HerBERT 模型)
• hubert — Wav2Vec2CTCTokenizer (Hubert 模型)
• ibert — RobertaTokenizer 或 RobertaTokenizerFast (I-BERT 模型)
• idefics — LlamaTokenizerFast (IDEFICS 模型)
• idefics2 — LlamaTokenizer 或 LlamaTokenizerFast (Idefics2 模型)
• idefics3 — LlamaTokenizer 或 LlamaTokenizerFast (Idefics3 模型)
• instructblip — GPT2Tokenizer 或 GPT2TokenizerFast (InstructBLIP 模型)
• instructblipvideo — GPT2Tokenizer 或 GPT2TokenizerFast (InstructBlipVideo 模型)
• jamba — LlamaTokenizer 或 LlamaTokenizerFast (Jamba 模型)
• jetmoe — LlamaTokenizer 或 LlamaTokenizerFast (JetMoe 模型)
• jukebox — JukeboxTokenizer (Jukebox 模型)
• kosmos-2 — XLMRobertaTokenizer 或 XLMRobertaTokenizerFast (KOSMOS-2 模型)
• layoutlm — LayoutLMTokenizer 或 LayoutLMTokenizerFast (LayoutLM 模型)
• layoutlmv2 — LayoutLMv2Tokenizer 或 LayoutLMv2TokenizerFast (LayoutLMv2 模型)
• layoutlmv3 — LayoutLMv3Tokenizer 或 LayoutLMv3TokenizerFast (LayoutLMv3 模型)
• layoutxlm — LayoutXLMTokenizer 或 LayoutXLMTokenizerFast (LayoutXLM 模型)
• led — LEDTokenizer 或 LEDTokenizerFast (LED 模型)
• lilt — LayoutLMv3Tokenizer 或 LayoutLMv3TokenizerFast (LiLT 模型)
• llama — LlamaTokenizer 或 LlamaTokenizerFast (LLaMA 模型)
• llava — LlamaTokenizer 或 LlamaTokenizerFast (LLaVa 模型)
• llava_next — LlamaTokenizer 或 LlamaTokenizerFast (LLaVA-NeXT 模型)
• llava_next_video — LlamaTokenizer 或 LlamaTokenizerFast (LLaVa-NeXT-Video 模型)
• llava_onevision — LlamaTokenizer 或 LlamaTokenizerFast (LLaVA-Onevision 模型)
• longformer — LongformerTokenizer 或 LongformerTokenizerFast (Longformer 模型)
• longt5 — T5Tokenizer 或 T5TokenizerFast (LongT5 模型)
• luke — LukeTokenizer (LUKE 模型)
• lxmert — LxmertTokenizer 或 LxmertTokenizerFast (LXMERT 模型)
• m2m_100 — M2M100Tokenizer (M2M100 模型)
• mamba — GPTNeoXTokenizerFast (Mamba 模型)
• mamba2 — GPTNeoXTokenizerFast (mamba2 模型)
• marian — MarianTokenizer (Marian 模型)
• mbart — MBartTokenizer 或 MBartTokenizerFast (mBART 模型)
• mbart50 — MBart50Tokenizer 或 MBart50TokenizerFast (mBART-50 模型)
• mega — RobertaTokenizer 或 RobertaTokenizerFast (MEGA 模型)
• megatron-bert — BertTokenizer 或 BertTokenizerFast (Megatron-BERT 模型)
• mgp-str — MgpstrTokenizer (MGP-STR 模型)
• mistral — LlamaTokenizer 或 LlamaTokenizerFast (Mistral 模型)
• mixtral — LlamaTokenizer 或 LlamaTokenizerFast (Mixtral 模型)
• mllama — LlamaTokenizer 或 LlamaTokenizerFast (Mllama 模型)
• mluke — MLukeTokenizer (mLUKE 模型)
• mobilebert — MobileBertTokenizer 或 MobileBertTokenizerFast (MobileBERT 模型)
• modernbert — PreTrainedTokenizerFast (ModernBERT 模型)
• moonshine — PreTrainedTokenizerFast (Moonshine 模型)
• moshi — PreTrainedTokenizerFast (Moshi 模型)
• mpnet — MPNetTokenizer 或 MPNetTokenizerFast (MPNet 模型)
• mpt — GPTNeoXTokenizerFast (MPT 模型)
• mra — RobertaTokenizer 或 RobertaTokenizerFast (MRA 模型)
• mt5 — MT5Tokenizer 或 MT5TokenizerFast (MT5 模型)
• musicgen — T5Tokenizer 或 T5TokenizerFast (MusicGen 模型)
• musicgen_melody — T5Tokenizer 或 T5TokenizerFast (MusicGen Melody 模型)
• mvp — MvpTokenizer 或 MvpTokenizerFast (MVP 模型)
• myt5 — MyT5Tokenizer (myt5 模型)
• nemotron — PreTrainedTokenizerFast (Nemotron 模型)
• nezha — BertTokenizer 或 BertTokenizerFast (Nezha 模型)
• nllb — NllbTokenizer 或 NllbTokenizerFast (NLLB 模型)
• nllb-moe — NllbTokenizer 或 NllbTokenizerFast (NLLB-MOE 模型)
• nystromformer — AlbertTokenizer 或 AlbertTokenizerFast (Nyströmformer 模型)
• olmo — GPTNeoXTokenizerFast (OLMo 模型)
• olmo2 — GPTNeoXTokenizerFast (OLMo2 模型)
• olmoe — GPTNeoXTokenizerFast (OLMoE 模型)
• omdet-turbo — CLIPTokenizer 或 CLIPTokenizerFast (OmDet-Turbo 模型)
• oneformer — CLIPTokenizer 或 CLIPTokenizerFast (OneFormer 模型)
• openai-gpt — OpenAIGPTTokenizer 或 OpenAIGPTTokenizerFast (OpenAI GPT 模型)
• opt — GPT2Tokenizer 或 GPT2TokenizerFast (OPT 模型)
• owlv2 — CLIPTokenizer 或 CLIPTokenizerFast (OWLv2 模型)
• owlvit — CLIPTokenizer 或 CLIPTokenizerFast (OWL-ViT 模型)
• paligemma — LlamaTokenizer 或 LlamaTokenizerFast (PaliGemma 模型)
• pegasus — PegasusTokenizer 或 PegasusTokenizerFast (Pegasus 模型)
• pegasus_x — PegasusTokenizer 或 PegasusTokenizerFast (PEGASUS-X 模型)
• perceiver — PerceiverTokenizer (Perceiver 模型)
• persimmon — LlamaTokenizer 或 LlamaTokenizerFast (Persimmon 模型)
• phi — CodeGenTokenizer 或 CodeGenTokenizerFast (Phi 模型)
• phi3 — LlamaTokenizer 或 LlamaTokenizerFast (Phi3 模型)
• phimoe — LlamaTokenizer 或 LlamaTokenizerFast (Phimoe 模型)
• phobert — PhobertTokenizer (PhoBERT 模型)
• pix2struct — T5Tokenizer 或 T5TokenizerFast (Pix2Struct 模型)
• pixtral — PreTrainedTokenizerFast (Pixtral 模型)
• plbart — PLBartTokenizer (PLBart 模型)
• prophetnet — ProphetNetTokenizer (ProphetNet 模型)
• qdqbert — BertTokenizer 或 BertTokenizerFast (QDQBert 模型)
• qwen2 — Qwen2Tokenizer 或 Qwen2TokenizerFast (Qwen2 模型)
• qwen2_5_vl — Qwen2Tokenizer 或 Qwen2TokenizerFast (Qwen2_5_VL 模型)
• qwen2_audio — Qwen2Tokenizer 或 Qwen2TokenizerFast (Qwen2Audio 模型)
• qwen2_moe — Qwen2Tokenizer 或 Qwen2TokenizerFast (Qwen2MoE 模型)
• qwen2_vl — Qwen2Tokenizer 或 Qwen2TokenizerFast (Qwen2VL 模型)
• rag — RagTokenizer (RAG 模型)
• realm — RealmTokenizer 或 RealmTokenizerFast (REALM 模型)
• recurrent_gemma — GemmaTokenizer 或 GemmaTokenizerFast (RecurrentGemma 模型)
• reformer — ReformerTokenizer 或 ReformerTokenizerFast (Reformer 模型)
• rembert — RemBertTokenizer 或 RemBertTokenizerFast (RemBERT 模型)
• retribert — RetriBertTokenizer 或 RetriBertTokenizerFast (RetriBERT 模型)
• roberta — RobertaTokenizer 或 RobertaTokenizerFast (RoBERTa 模型)
• roberta-prelayernorm — RobertaTokenizer 或 RobertaTokenizerFast (RoBERTa-PreLayerNorm 模型)
• roc_bert — RoCBertTokenizer (RoCBert 模型)
• roformer — RoFormerTokenizer 或 RoFormerTokenizerFast (RoFormer 模型)
• rwkv — GPTNeoXTokenizerFast (RWKV 模型)
• seamless_m4t — SeamlessM4TTokenizer 或 SeamlessM4TTokenizerFast (SeamlessM4T 模型)
• seamless_m4t_v2 — SeamlessM4TTokenizer 或 SeamlessM4TTokenizerFast (SeamlessM4Tv2 模型)
• shieldgemma2 — GemmaTokenizer 或 GemmaTokenizerFast (Shieldgemma2 模型)
• siglip — SiglipTokenizer (SigLIP 模型)
• siglip2 — GemmaTokenizer 或 GemmaTokenizerFast (SigLIP2 模型)
• speech_to_text — Speech2TextTokenizer (Speech2Text 模型)
• speech_to_text_2 — Speech2Text2Tokenizer (Speech2Text2 模型)
• speecht5 — SpeechT5Tokenizer (SpeechT5 模型)
• splinter — SplinterTokenizer 或 SplinterTokenizerFast (Splinter 模型)
• squeezebert — SqueezeBertTokenizer 或 SqueezeBertTokenizerFast (SqueezeBERT 模型)
• stablelm — GPTNeoXTokenizerFast (StableLm 模型)
• starcoder2 — GPT2Tokenizer 或 GPT2TokenizerFast (Starcoder2 模型)
• switch_transformers — T5Tokenizer 或 T5TokenizerFast (SwitchTransformers 模型)
• t5 — T5Tokenizer 或 T5TokenizerFast (T5 模型)
• tapas — TapasTokenizer (TAPAS 模型)
• tapex — TapexTokenizer (TAPEX 模型)
• transfo-xl — TransfoXLTokenizer (Transformer-XL 模型)
• tvp — BertTokenizer 或 BertTokenizerFast (TVP 模型)
• udop — UdopTokenizer 或 UdopTokenizerFast (UDOP 模型)
• umt5 — T5Tokenizer 或 T5TokenizerFast (UMT5 模型)
• video_llava — LlamaTokenizer 或 LlamaTokenizerFast (VideoLlava 模型)
• vilt — BertTokenizer 或 BertTokenizerFast (ViLT 模型)
• vipllava — LlamaTokenizer 或 LlamaTokenizerFast (VipLlava 模型)
• visual_bert — BertTokenizer 或 BertTokenizerFast (VisualBERT 模型)
• vits — VitsTokenizer (VITS 模型)
• wav2vec2 — Wav2Vec2CTCTokenizer (Wav2Vec2 模型)
• wav2vec2-bert — Wav2Vec2CTCTokenizer (Wav2Vec2-BERT 模型)
• wav2vec2-conformer — Wav2Vec2CTCTokenizer (Wav2Vec2-Conformer 模型)
• wav2vec2_phoneme — Wav2Vec2PhonemeCTCTokenizer (Wav2Vec2Phoneme 模型)
• whisper — WhisperTokenizer 或 WhisperTokenizerFast (Whisper 模型)
• xclip — CLIPTokenizer 或 CLIPTokenizerFast (X-CLIP 模型)
• xglm — XGLMTokenizer 或 XGLMTokenizerFast (XGLM 模型)
• xlm — XLMTokenizer (XLM 模型)
• xlm-prophetnet — XLMProphetNetTokenizer (XLM-ProphetNet 模型)
• xlm-roberta — XLMRobertaTokenizer 或 XLMRobertaTokenizerFast (XLM-RoBERTa 模型)
• xlm-roberta-xl — XLMRobertaTokenizer 或 XLMRobertaTokenizerFast (XLM-RoBERTa-XL 模型)
• xlnet — XLNetTokenizer 或 XLNetTokenizerFast (XLNet 模型)
• xmod — XLMRobertaTokenizer 或 XLMRobertaTokenizerFast (X-MOD 模型)
• yoso — AlbertTokenizer 或 AlbertTokenizerFast (YOSO 模型)
• zamba — LlamaTokenizer 或 LlamaTokenizerFast (Zamba 模型)
• zamba2 — LlamaTokenizer 或 LlamaTokenizerFast (Zamba2 模型)

>>> from transformers import AutoTokenizer

>>> # Download vocabulary from huggingface.co and cache.
>>> tokenizer = AutoTokenizer.from_pretrained("google-bert/bert-base-uncased")

>>> # Download vocabulary from huggingface.co (user-uploaded) and cache.
>>> tokenizer = AutoTokenizer.from_pretrained("dbmdz/bert-base-german-cased")

>>> # If vocabulary files are in a directory (e.g. tokenizer was saved using *save_pretrained('./test/saved_model/')*)
>>> # tokenizer = AutoTokenizer.from_pretrained("./test/bert_saved_model/")

>>> # Download vocabulary from huggingface.co and define model-specific arguments
>>> tokenizer = AutoTokenizer.from_pretrained("FacebookAI/roberta-base", add_prefix_space=True)

在此映射中注册一个新的分词器。

从预训练模型词汇表实例化库中的特征提取器类之一。

要实例化的特征提取器类是基于配置对象的 model_type 属性（作为参数传递或从 pretrained_model_name_or_path 加载，如果可能）选择的，或者当它缺失时，通过回退到使用 pretrained_model_name_or_path 上的模式匹配来选择。

• audio-spectrogram-transformer — ASTFeatureExtractor (音频频谱图 Transformer 模型)
• beit — BeitFeatureExtractor (BEiT 模型)
• chinese_clip — ChineseCLIPFeatureExtractor (Chinese-CLIP 模型)
• clap — ClapFeatureExtractor (CLAP 模型)
• clip — CLIPFeatureExtractor (CLIP 模型)
• clipseg — ViTFeatureExtractor (CLIPSeg 模型)
• clvp — ClvpFeatureExtractor (CLVP 模型)
• conditional_detr — ConditionalDetrFeatureExtractor (Conditional DETR 模型)
• convnext — ConvNextFeatureExtractor (ConvNeXT 模型)
• cvt — ConvNextFeatureExtractor (CvT 模型)
• dac — DacFeatureExtractor (DAC 模型)
• data2vec-audio — Wav2Vec2FeatureExtractor (Data2VecAudio 模型)
• data2vec-vision — BeitFeatureExtractor (Data2VecVision 模型)
• deformable_detr — DeformableDetrFeatureExtractor (Deformable DETR 模型)
• deit — DeiTFeatureExtractor (DeiT 模型)
• detr — DetrFeatureExtractor (DETR 模型)
• dinat — ViTFeatureExtractor (DiNAT 模型)
• donut-swin — DonutFeatureExtractor (DonutSwin 模型)
• dpt — DPTFeatureExtractor (DPT 模型)
• encodec — EncodecFeatureExtractor (EnCodec 模型)
• flava — FlavaFeatureExtractor (FLAVA 模型)
• glpn — GLPNFeatureExtractor (GLPN 模型)
• groupvit — CLIPFeatureExtractor (GroupViT 模型)
• hubert — Wav2Vec2FeatureExtractor (Hubert 模型)
• imagegpt — ImageGPTFeatureExtractor (ImageGPT 模型)
• layoutlmv2 — LayoutLMv2FeatureExtractor (LayoutLMv2 模型)
• layoutlmv3 — LayoutLMv3FeatureExtractor (LayoutLMv3 模型)
• levit — LevitFeatureExtractor (LeViT 模型)
• maskformer — MaskFormerFeatureExtractor (MaskFormer 模型)
• mctct — MCTCTFeatureExtractor (M-CTC-T 模型)
• mimi — EncodecFeatureExtractor (Mimi 模型)
• mobilenet_v1 — MobileNetV1FeatureExtractor (MobileNetV1 模型)
• mobilenet_v2 — MobileNetV2FeatureExtractor (MobileNetV2 模型)
• mobilevit — MobileViTFeatureExtractor (MobileViT 模型)
• moonshine — Wav2Vec2FeatureExtractor (Moonshine 模型)
• moshi — EncodecFeatureExtractor (Moshi 模型)
• nat — ViTFeatureExtractor (NAT 模型)
• owlvit — OwlViTFeatureExtractor (OWL-ViT 模型)
• perceiver — PerceiverFeatureExtractor (Perceiver 模型)
• poolformer — PoolFormerFeatureExtractor (PoolFormer 模型)
• pop2piano — Pop2PianoFeatureExtractor (Pop2Piano 模型)
• regnet — ConvNextFeatureExtractor (RegNet 模型)
• resnet — ConvNextFeatureExtractor (ResNet 模型)
• seamless_m4t — SeamlessM4TFeatureExtractor (SeamlessM4T 模型)
• seamless_m4t_v2 — SeamlessM4TFeatureExtractor (SeamlessM4Tv2 模型)
• segformer — SegformerFeatureExtractor (SegFormer 模型)
• sew — Wav2Vec2FeatureExtractor (SEW 模型)
• sew-d — Wav2Vec2FeatureExtractor (SEW-D 模型)
• speech_to_text — Speech2TextFeatureExtractor (Speech2Text 模型)
• speecht5 — SpeechT5FeatureExtractor (SpeechT5 模型)
• swiftformer — ViTFeatureExtractor (SwiftFormer 模型)
• swin — ViTFeatureExtractor (Swin Transformer 模型)
• swinv2 — ViTFeatureExtractor (Swin Transformer V2 模型)
• table-transformer — DetrFeatureExtractor (Table Transformer 模型)
• timesformer — VideoMAEFeatureExtractor (TimeSformer 模型)
• tvlt — TvltFeatureExtractor (TVLT 模型)
• unispeech — Wav2Vec2FeatureExtractor (UniSpeech 模型)
• unispeech-sat — Wav2Vec2FeatureExtractor (UniSpeechSat 模型)
• univnet — UnivNetFeatureExtractor (UnivNet 模型)
• van — ConvNextFeatureExtractor (VAN 模型)
• videomae — VideoMAEFeatureExtractor (VideoMAE 模型)
• vilt — ViltFeatureExtractor (ViLT 模型)
• vit — ViTFeatureExtractor (ViT 模型)
• vit_mae — ViTFeatureExtractor (ViTMAE 模型)
• vit_msn — ViTFeatureExtractor (ViTMSN 模型)
• wav2vec2 — Wav2Vec2FeatureExtractor (Wav2Vec2 模型)
• wav2vec2-bert — Wav2Vec2FeatureExtractor (Wav2Vec2-BERT 模型)
• wav2vec2-conformer — Wav2Vec2FeatureExtractor (Wav2Vec2-Conformer 模型)
• wavlm — Wav2Vec2FeatureExtractor (WavLM 模型)
• whisper — WhisperFeatureExtractor (Whisper 模型)
• xclip — CLIPFeatureExtractor (X-CLIP 模型)
• yolos — YolosFeatureExtractor (YOLOS 模型)

>>> from transformers import AutoFeatureExtractor

>>> # Download feature extractor from huggingface.co and cache.
>>> feature_extractor = AutoFeatureExtractor.from_pretrained("facebook/wav2vec2-base-960h")

>>> # If feature extractor files are in a directory (e.g. feature extractor was saved using *save_pretrained('./test/saved_model/')*)
>>> # feature_extractor = AutoFeatureExtractor.from_pretrained("./test/saved_model/")

为此类注册一个新的特征提取器。

从预训练模型词汇表实例化库中的一个图像处理器类。

要实例化的图像处理器类是基于配置对象的 model_type 属性（作为参数传递或从 pretrained_model_name_or_path 加载，如果可能）选择的，或者在缺少 model_type 属性时，通过回退到在 pretrained_model_name_or_path 上使用模式匹配来选择。

• align — EfficientNetImageProcessor (ALIGN 模型)
• aria — A 或 r (Aria 模型)
• beit — BeitImageProcessor (BEiT 模型)
• bit — BitImageProcessor (BiT 模型)
• blip — BlipImageProcessor 或 BlipImageProcessorFast (BLIP 模型)
• blip-2 — BlipImageProcessor 或 BlipImageProcessorFast (BLIP-2 模型)
• bridgetower — BridgeTowerImageProcessor (BridgeTower 模型)
• chameleon — ChameleonImageProcessor (Chameleon 模型)
• chinese_clip — ChineseCLIPImageProcessor (Chinese-CLIP 模型)
• clip — CLIPImageProcessor 或 CLIPImageProcessorFast (CLIP 模型)
• clipseg — ViTImageProcessor 或 ViTImageProcessorFast (CLIPSeg 模型)
• conditional_detr — ConditionalDetrImageProcessor (Conditional DETR 模型)
• convnext — ConvNextImageProcessor 或 ConvNextImageProcessorFast (ConvNeXT 模型)
• convnextv2 — ConvNextImageProcessor 或 ConvNextImageProcessorFast (ConvNeXTV2 模型)
• cvt — ConvNextImageProcessor 或 ConvNextImageProcessorFast (CvT 模型)
• data2vec-vision — BeitImageProcessor (Data2VecVision 模型)
• deformable_detr — DeformableDetrImageProcessor 或 DeformableDetrImageProcessorFast (Deformable DETR 模型)
• deit — DeiTImageProcessor 或 DeiTImageProcessorFast (DeiT 模型)
• depth_anything — DPTImageProcessor (Depth Anything 模型)
• depth_pro — DepthProImageProcessor 或 DepthProImageProcessorFast (DepthPro 模型)
• deta — DetaImageProcessor (DETA 模型)
• detr — DetrImageProcessor 或 DetrImageProcessorFast (DETR 模型)
• dinat — ViTImageProcessor 或 ViTImageProcessorFast (DiNAT 模型)
• dinov2 — BitImageProcessor (DINOv2 模型)
• donut-swin — DonutImageProcessor (DonutSwin 模型)
• dpt — DPTImageProcessor (DPT 模型)
• efficientformer — EfficientFormerImageProcessor (EfficientFormer 模型)
• efficientnet — EfficientNetImageProcessor (EfficientNet 模型)
• flava — FlavaImageProcessor (FLAVA 模型)
• focalnet — BitImageProcessor (FocalNet 模型)
• fuyu — FuyuImageProcessor (Fuyu 模型)
• gemma3 — Gemma3ImageProcessor 或 Gemma3ImageProcessorFast (Gemma3ForConditionalGeneration 模型)
• git — CLIPImageProcessor 或 CLIPImageProcessorFast (GIT 模型)
• glpn — GLPNImageProcessor (GLPN 模型)
• got_ocr2 — GotOcr2ImageProcessor 或 GotOcr2ImageProcessorFast (GOT-OCR2 模型)
• grounding-dino — GroundingDinoImageProcessor (Grounding DINO 模型)
• groupvit — CLIPImageProcessor 或 CLIPImageProcessorFast (GroupViT 模型)
• hiera — BitImageProcessor (Hiera 模型)
• idefics — IdeficsImageProcessor (IDEFICS 模型)
• idefics2 — Idefics2ImageProcessor (Idefics2 模型)
• idefics3 — Idefics3ImageProcessor (Idefics3 模型)
• ijepa — ViTImageProcessor 或 ViTImageProcessorFast (I-JEPA 模型)
• imagegpt — ImageGPTImageProcessor (ImageGPT 模型)
• instructblip — BlipImageProcessor 或 BlipImageProcessorFast (InstructBLIP 模型)
• instructblipvideo — InstructBlipVideoImageProcessor (InstructBlipVideo 模型)
• kosmos-2 — CLIPImageProcessor 或 CLIPImageProcessorFast (KOSMOS-2 模型)
• layoutlmv2 — LayoutLMv2ImageProcessor (LayoutLMv2 模型)
• layoutlmv3 — LayoutLMv3ImageProcessor (LayoutLMv3 模型)
• levit — LevitImageProcessor (LeViT 模型)
• llava — LlavaImageProcessor 或 LlavaImageProcessorFast (LLaVa 模型)
• llava_next — LlavaNextImageProcessor 或 LlavaNextImageProcessorFast (LLaVA-NeXT 模型)
• llava_next_video — LlavaNextVideoImageProcessor (LLaVa-NeXT-Video 模型)
• llava_onevision — LlavaOnevisionImageProcessor 或 LlavaOnevisionImageProcessorFast (LLaVA-Onevision 模型)
• mask2former — Mask2FormerImageProcessor (Mask2Former 模型)
• maskformer — MaskFormerImageProcessor (MaskFormer 模型)
• mgp-str — ViTImageProcessor 或 ViTImageProcessorFast (MGP-STR 模型)
• mistral3 — PixtralImageProcessor 或 PixtralImageProcessorFast (Mistral3 模型)
• mllama — MllamaImageProcessor (Mllama 模型)
• mobilenet_v1 — MobileNetV1ImageProcessor (MobileNetV1 模型)
• mobilenet_v2 — MobileNetV2ImageProcessor (MobileNetV2 模型)
• mobilevit — MobileViTImageProcessor (MobileViT 模型)
• mobilevitv2 — MobileViTImageProcessor (MobileViTV2 模型)
• nat — ViTImageProcessor 或 ViTImageProcessorFast (NAT 模型)
• nougat — NougatImageProcessor (Nougat 模型)
• oneformer — OneFormerImageProcessor (OneFormer 模型)
• owlv2 — Owlv2ImageProcessor (OWLv2 模型)
• owlvit — OwlViTImageProcessor (OWL-ViT 模型)
• paligemma — SiglipImageProcessor 或 SiglipImageProcessorFast (PaliGemma 模型)
• perceiver — PerceiverImageProcessor (Perceiver 模型)
• pix2struct — Pix2StructImageProcessor (Pix2Struct 模型)
• pixtral — PixtralImageProcessor 或 PixtralImageProcessorFast (Pixtral 模型)
• poolformer — PoolFormerImageProcessor (PoolFormer 模型)
• prompt_depth_anything — PromptDepthAnythingImageProcessor (PromptDepthAnything 模型)
• pvt — PvtImageProcessor (PVT 模型)
• pvt_v2 — PvtImageProcessor (PVTv2 模型)
• qwen2_5_vl — Qwen2VLImageProcessor 或 Qwen2VLImageProcessorFast (Qwen2_5_VL 模型)
• qwen2_vl — Qwen2VLImageProcessor 或 Qwen2VLImageProcessorFast (Qwen2VL 模型)
• regnet — ConvNextImageProcessor 或 ConvNextImageProcessorFast (RegNet 模型)
• resnet — ConvNextImageProcessor 或 ConvNextImageProcessorFast (ResNet 模型)
• rt_detr — RTDetrImageProcessor 或 RTDetrImageProcessorFast (RT-DETR 模型)
• sam — SamImageProcessor (SAM 模型)
• segformer — SegformerImageProcessor (SegFormer 模型)
• seggpt — SegGptImageProcessor (SegGPT 模型)
• shieldgemma2 — Gemma3ImageProcessor 或 Gemma3ImageProcessorFast (Shieldgemma2 模型)
• siglip — SiglipImageProcessor 或 SiglipImageProcessorFast (SigLIP 模型)
• siglip2 — Siglip2ImageProcessor 或 Siglip2ImageProcessorFast (SigLIP2 模型)
• superglue — S 或 u (SuperGlue 模型)
• swiftformer — ViTImageProcessor 或 ViTImageProcessorFast (SwiftFormer 模型)
• swin — ViTImageProcessor 或 ViTImageProcessorFast (Swin Transformer 模型)
• swin2sr — Swin2SRImageProcessor (Swin2SR 模型)
• swinv2 — ViTImageProcessor 或 ViTImageProcessorFast (Swin Transformer V2 模型)
• table-transformer — DetrImageProcessor (Table Transformer 模型)
• timesformer — VideoMAEImageProcessor (TimeSformer 模型)
• timm_wrapper — TimmWrapperImageProcessor (TimmWrapperModel 模型)
• tvlt — TvltImageProcessor (TVLT 模型)
• tvp — TvpImageProcessor (TVP 模型)
• udop — LayoutLMv3ImageProcessor (UDOP 模型)
• upernet — SegformerImageProcessor (UPerNet 模型)
• van — ConvNextImageProcessor 或 ConvNextImageProcessorFast (VAN 模型)
• videomae — VideoMAEImageProcessor (VideoMAE 模型)
• vilt — ViltImageProcessor (ViLT 模型)
• vipllava — CLIPImageProcessor 或 CLIPImageProcessorFast (VipLlava 模型)
• vit — ViTImageProcessor 或 ViTImageProcessorFast (ViT 模型)
• vit_hybrid — ViTHybridImageProcessor (ViT Hybrid 模型)
• vit_mae — ViTImageProcessor 或 ViTImageProcessorFast (ViTMAE 模型)
• vit_msn — ViTImageProcessor 或 ViTImageProcessorFast (ViTMSN 模型)
• vitmatte — VitMatteImageProcessor (ViTMatte 模型)
• xclip — CLIPImageProcessor 或 CLIPImageProcessorFast (X-CLIP 模型)
• yolos — YolosImageProcessor (YOLOS 模型)
• zoedepth — ZoeDepthImageProcessor (ZoeDepth 模型)

>>> from transformers import AutoImageProcessor

>>> # Download image processor from huggingface.co and cache.
>>> image_processor = AutoImageProcessor.from_pretrained("google/vit-base-patch16-224-in21k")

>>> # If image processor files are in a directory (e.g. image processor was saved using *save_pretrained('./test/saved_model/')*)
>>> # image_processor = AutoImageProcessor.from_pretrained("./test/saved_model/")

为此类注册一个新的图像处理器。

从预训练模型词汇表实例化库中的一个处理器类。

要实例化的处理器类是根据配置对象的 model_type 属性选择的（可以作为参数传递，也可以从 pretrained_model_name_or_path 加载，如果可能）。

• align — AlignProcessor (ALIGN 模型)
• altclip — AltCLIPProcessor (AltCLIP 模型)
• aria — AriaProcessor (Aria 模型)
• aya_vision — AyaVisionProcessor (AyaVision 模型)
• bark — BarkProcessor (Bark 模型)
• blip — BlipProcessor (BLIP 模型)
• blip-2 — Blip2Processor (BLIP-2 模型)
• bridgetower — BridgeTowerProcessor (BridgeTower 模型)
• chameleon — ChameleonProcessor (Chameleon 模型)
• chinese_clip — ChineseCLIPProcessor (Chinese-CLIP 模型)
• clap — ClapProcessor (CLAP 模型)
• clip — CLIPProcessor (CLIP 模型)
• clipseg — CLIPSegProcessor (CLIPSeg 模型)
• clvp — ClvpProcessor (CLVP 模型)
• colpali — ColPaliProcessor (ColPali 模型)
• emu3 — Emu3Processor (Emu3 模型)
• flava — FlavaProcessor (FLAVA 模型)
• fuyu — FuyuProcessor (Fuyu 模型)
• gemma3 — Gemma3Processor (Gemma3ForConditionalGeneration 模型)
• git — GitProcessor (GIT 模型)
• got_ocr2 — GotOcr2Processor (GOT-OCR2 模型)
• grounding-dino — GroundingDinoProcessor (Grounding DINO 模型)
• groupvit — CLIPProcessor (GroupViT 模型)
• hubert — Wav2Vec2Processor (Hubert 模型)
• idefics — IdeficsProcessor (IDEFICS 模型)
• idefics2 — Idefics2Processor (Idefics2 模型)
• idefics3 — Idefics3Processor (Idefics3 模型)
• instructblip — InstructBlipProcessor (InstructBLIP 模型)
• instructblipvideo — InstructBlipVideoProcessor (InstructBlipVideo 模型)
• kosmos-2 — Kosmos2Processor (KOSMOS-2 模型)
• layoutlmv2 — LayoutLMv2Processor (LayoutLMv2 模型)
• layoutlmv3 — LayoutLMv3Processor (LayoutLMv3 模型)
• llava — LlavaProcessor (LLaVa 模型)
• llava_next — LlavaNextProcessor (LLaVA-NeXT 模型)
• llava_next_video — LlavaNextVideoProcessor (LLaVa-NeXT-Video 模型)
• llava_onevision — LlavaOnevisionProcessor (LLaVA-Onevision 模型)
• markuplm — MarkupLMProcessor (MarkupLM 模型)
• mctct — MCTCTProcessor (M-CTC-T 模型)
• mgp-str — MgpstrProcessor (MGP-STR 模型)
• mistral3 — PixtralProcessor (Mistral3 模型)
• mllama — MllamaProcessor (Mllama 模型)
• moonshine — Wav2Vec2Processor (Moonshine 模型)
• oneformer — OneFormerProcessor (OneFormer 模型)
• owlv2 — Owlv2Processor (OWLv2 模型)
• owlvit — OwlViTProcessor (OWL-ViT 模型)
• paligemma — PaliGemmaProcessor (PaliGemma 模型)
• pix2struct — Pix2StructProcessor (Pix2Struct 模型)
• pixtral — PixtralProcessor (Pixtral 模型)
• pop2piano — Pop2PianoProcessor (Pop2Piano 模型)
• qwen2_5_vl — Qwen2_5_VLProcessor (Qwen2_5_VL 模型)
• qwen2_audio — Qwen2AudioProcessor (Qwen2Audio 模型)
• qwen2_vl — Qwen2VLProcessor (Qwen2VL 模型)
• sam — SamProcessor (SAM 模型)
• seamless_m4t — SeamlessM4TProcessor (SeamlessM4T 模型)
• sew — Wav2Vec2Processor (SEW 模型)
• sew-d — Wav2Vec2Processor (SEW-D 模型)
• shieldgemma2 — ShieldGemma2Processor (Shieldgemma2 模型)
• siglip — SiglipProcessor (SigLIP 模型)
• siglip2 — Siglip2Processor (SigLIP2 模型)
• speech_to_text — Speech2TextProcessor (Speech2Text 模型)
• speech_to_text_2 — Speech2Text2Processor (Speech2Text2 模型)
• speecht5 — SpeechT5Processor (SpeechT5 模型)
• trocr — TrOCRProcessor (TrOCR 模型)
• tvlt — TvltProcessor (TVLT 模型)
• tvp — TvpProcessor (TVP 模型)
• udop — UdopProcessor (UDOP 模型)
• unispeech — Wav2Vec2Processor (UniSpeech 模型)
• unispeech-sat — Wav2Vec2Processor (UniSpeechSat 模型)
• video_llava — VideoLlavaProcessor (VideoLlava 模型)
• vilt — ViltProcessor (ViLT 模型)
• vipllava — LlavaProcessor (VipLlava 模型)
• vision-text-dual-encoder — VisionTextDualEncoderProcessor (VisionTextDualEncoder 模型)
• wav2vec2 — Wav2Vec2Processor (Wav2Vec2 模型)
• wav2vec2-bert — Wav2Vec2Processor (Wav2Vec2-BERT 模型)
• wav2vec2-conformer — Wav2Vec2Processor (Wav2Vec2-Conformer 模型)
• wavlm — Wav2Vec2Processor (WavLM 模型)
• whisper — WhisperProcessor (Whisper 模型)
• xclip — XCLIPProcessor (X-CLIP 模型)

>>> from transformers import AutoProcessor

>>> # Download processor from huggingface.co and cache.
>>> processor = AutoProcessor.from_pretrained("facebook/wav2vec2-base-960h")

>>> # If processor files are in a directory (e.g. processor was saved using *save_pretrained('./test/saved_model/')*)
>>> # processor = AutoProcessor.from_pretrained("./test/saved_model/")

为此类注册一个新的处理器。

从配置实例化库的基础模型类之一。

注意： 从其配置文件加载模型**不会**加载模型权重。它仅影响模型的配置。使用 from_pretrained() 加载模型权重。

>>> from transformers import AutoConfig, AutoModel

>>> # Download configuration from huggingface.co and cache.
>>> config = AutoConfig.from_pretrained("google-bert/bert-base-cased")
>>> model = AutoModel.from_config(config)

从预训练模型实例化库的基础模型类之一。

要实例化的模型类是根据配置对象的 model_type 属性（作为参数传递或从 pretrained_model_name_or_path 加载，如果可能）选择的，或者当它丢失时，通过回退到在 pretrained_model_name_or_path 上使用模式匹配来选择。

• albert — AlbertModel (ALBERT 模型)
• align — AlignModel (ALIGN 模型)
• altclip — AltCLIPModel (AltCLIP 模型)
• aria — AriaForConditionalGeneration (Aria 模型)
• aria_text — AriaTextModel (AriaText 模型)
• audio-spectrogram-transformer — ASTModel (音频频谱图转换器模型)
• autoformer — AutoformerModel (Autoformer 模型)
• bamba — BambaModel (Bamba 模型)
• bark — BarkModel (Bark 模型)
• bart — BartModel (BART 模型)
• beit — BeitModel (BEiT 模型)
• bert — BertModel (BERT 模型)
• bert-generation — BertGenerationEncoder (Bert Generation 模型)
• big_bird — BigBirdModel (BigBird 模型)
• bigbird_pegasus — BigBirdPegasusModel (BigBird-Pegasus 模型)
• biogpt — BioGptModel (BioGpt 模型)
• bit — BitModel (BiT 模型)
• blenderbot — BlenderbotModel (Blenderbot 模型)
• blenderbot-small — BlenderbotSmallModel (BlenderbotSmall 模型)
• blip — BlipModel (BLIP 模型)
• blip-2 — Blip2Model (BLIP-2 模型)
• bloom — BloomModel (BLOOM 模型)
• bridgetower — BridgeTowerModel (BridgeTower 模型)
• bros — BrosModel (BROS 模型)
• camembert — CamembertModel (CamemBERT 模型)
• canine — CanineModel (CANINE 模型)
• chameleon — ChameleonModel (Chameleon 模型)
• chinese_clip — ChineseCLIPModel (Chinese-CLIP 模型)
• chinese_clip_vision_model — ChineseCLIPVisionModel (ChineseCLIPVisionModel 模型)
• clap — ClapModel (CLAP 模型)
• clip — CLIPModel (CLIP 模型)
• clip_text_model — CLIPTextModel (CLIPTextModel 模型)
• clip_vision_model — CLIPVisionModel (CLIPVisionModel 模型)
• clipseg — CLIPSegModel (CLIPSeg 模型)
• clvp — ClvpModelForConditionalGeneration (CLVP 模型)
• code_llama — LlamaModel (CodeLlama 模型)
• codegen — CodeGenModel (CodeGen 模型)
• cohere — CohereModel (Cohere 模型)
• cohere2 — Cohere2Model (Cohere2 模型)
• conditional_detr — ConditionalDetrModel (Conditional DETR 模型)
• convbert — ConvBertModel (ConvBERT 模型)
• convnext — ConvNextModel (ConvNeXT 模型)
• convnextv2 — ConvNextV2Model (ConvNeXTV2 模型)
• cpmant — CpmAntModel (CPM-Ant 模型)
• ctrl — CTRLModel (CTRL 模型)
• cvt — CvtModel (CvT 模型)
• dab-detr — DabDetrModel (DAB-DETR 模型)
• dac — DacModel (DAC 模型)
• data2vec-audio — Data2VecAudioModel (Data2VecAudio 模型)
• data2vec-text — Data2VecTextModel (Data2VecText 模型)
• data2vec-vision — Data2VecVisionModel (Data2VecVision 模型)
• dbrx — DbrxModel (DBRX 模型)
• deberta — DebertaModel (DeBERTa 模型)
• deberta-v2 — DebertaV2Model (DeBERTa-v2 模型)
• decision_transformer — DecisionTransformerModel (Decision Transformer 模型)
• deformable_detr — DeformableDetrModel (Deformable DETR 模型)
• deit — DeiTModel (DeiT 模型)
• depth_pro — DepthProModel (DepthPro 模型)
• deta — DetaModel (DETA 模型)
• detr — DetrModel (DETR 模型)
• diffllama — DiffLlamaModel (DiffLlama 模型)
• dinat — DinatModel (DiNAT 模型)
• dinov2 — Dinov2Model (DINOv2 模型)
• dinov2_with_registers — Dinov2WithRegistersModel (带寄存器的 DINOv2 模型)
• distilbert — DistilBertModel (DistilBERT 模型)
• donut-swin — DonutSwinModel (DonutSwin 模型)
• dpr — DPRQuestionEncoder (DPR 模型)
• dpt — DPTModel (DPT 模型)
• efficientformer — EfficientFormerModel (EfficientFormer 模型)
• efficientnet — EfficientNetModel (EfficientNet 模型)
• electra — ElectraModel (ELECTRA 模型)
• encodec — EncodecModel (EnCodec 模型)
• ernie — ErnieModel (ERNIE 模型)
• ernie_m — ErnieMModel (ErnieM 模型)
• esm — EsmModel (ESM 模型)
• falcon — FalconModel (Falcon 模型)
• falcon_mamba — FalconMambaModel (FalconMamba 模型)
• fastspeech2_conformer — FastSpeech2ConformerModel (FastSpeech2Conformer 模型)
• flaubert — FlaubertModel (FlauBERT 模型)
• flava — FlavaModel (FLAVA 模型)
• fnet — FNetModel (FNet 模型)
• focalnet — FocalNetModel (FocalNet 模型)
• fsmt — FSMTModel (FairSeq 机器翻译模型)
• funnel — FunnelModel 或 FunnelBaseModel (Funnel Transformer 模型)
• gemma — GemmaModel (Gemma 模型)
• gemma2 — Gemma2Model (Gemma2 模型)
• gemma3_text — Gemma3TextModel (Gemma3ForCausalLM 模型)
• git — GitModel (GIT 模型)
• glm — GlmModel (GLM 模型)
• glpn — GLPNModel (GLPN 模型)
• got_ocr2 — GotOcr2ForConditionalGeneration (GOT-OCR2 模型)
• gpt-sw3 — GPT2Model (GPT-Sw3 模型)
• gpt2 — GPT2Model (OpenAI GPT-2 模型)
• gpt_bigcode — GPTBigCodeModel (GPTBigCode 模型)
• gpt_neo — GPTNeoModel (GPT Neo 模型)
• gpt_neox — GPTNeoXModel (GPT NeoX 模型)
• gpt_neox_japanese — GPTNeoXJapaneseModel (GPT NeoX Japanese 模型)
• gptj — GPTJModel (GPT-J 模型)
• gptsan-japanese — GPTSanJapaneseForConditionalGeneration (GPTSAN-japanese 模型)
• granite — GraniteModel (Granite 模型)
• granitemoe — GraniteMoeModel (GraniteMoeMoe 模型)
• granitemoeshared — GraniteMoeSharedModel (GraniteMoeSharedMoe 模型)
• graphormer — GraphormerModel (Graphormer 模型)
• grounding-dino — GroundingDinoModel (Grounding DINO 模型)
• groupvit — GroupViTModel (GroupViT 模型)
• helium — HeliumModel (Helium 模型)
• hiera — HieraModel (Hiera 模型)
• hubert — HubertModel (Hubert 模型)
• ibert — IBertModel (I-BERT 模型)
• idefics — IdeficsModel (IDEFICS 模型)
• idefics2 — Idefics2Model (Idefics2 模型)
• idefics3 — Idefics3Model (Idefics3 模型)
• idefics3_vision — Idefics3VisionTransformer (Idefics3VisionTransformer 模型)
• ijepa — IJepaModel (I-JEPA 模型)
• imagegpt — ImageGPTModel (ImageGPT 模型)
• informer — InformerModel (Informer 模型)
• jamba — JambaModel (Jamba 模型)
• jetmoe — JetMoeModel (JetMoe 模型)
• jukebox — JukeboxModel (Jukebox 模型)
• kosmos-2 — Kosmos2Model (KOSMOS-2 模型)
• layoutlm — LayoutLMModel (LayoutLM 模型)
• layoutlmv2 — LayoutLMv2Model (LayoutLMv2 模型)
• layoutlmv3 — LayoutLMv3Model (LayoutLMv3 模型)
• led — LEDModel (LED 模型)
• levit — LevitModel (LeViT 模型)
• lilt — LiltModel (LiLT 模型)
• llama — LlamaModel (LLaMA 模型)
• longformer — LongformerModel (Longformer 模型)
• longt5 — LongT5Model (LongT5 模型)
• luke — LukeModel (LUKE 模型)
• lxmert — LxmertModel (LXMERT 模型)
• m2m_100 — M2M100Model (M2M100 模型)
• mamba — MambaModel (Mamba 模型)
• mamba2 — Mamba2Model (mamba2 模型)
• marian — MarianModel (Marian 模型)
• markuplm — MarkupLMModel (MarkupLM 模型)
• mask2former — Mask2FormerModel (Mask2Former 模型)
• maskformer — MaskFormerModel (MaskFormer 模型)
• maskformer-swin — MaskFormerSwinModel (MaskFormerSwin 模型)
• mbart — MBartModel (mBART 模型)
• mctct — MCTCTModel (M-CTC-T 模型)
• mega — MegaModel (MEGA 模型)
• megatron-bert — MegatronBertModel (Megatron-BERT 模型)
• mgp-str — MgpstrForSceneTextRecognition (MGP-STR 模型)
• mimi — MimiModel (Mimi 模型)
• mistral — MistralModel (Mistral 模型)
• mixtral — MixtralModel (Mixtral 模型)
• mobilebert — MobileBertModel (MobileBERT 模型)
• mobilenet_v1 — MobileNetV1Model (MobileNetV1 模型)
• mobilenet_v2 — MobileNetV2Model (MobileNetV2 模型)
• mobilevit — MobileViTModel (MobileViT 模型)
• mobilevitv2 — MobileViTV2Model (MobileViTV2 模型)
• modernbert — ModernBertModel (ModernBERT 模型)
• moonshine — MoonshineModel (Moonshine 模型)
• moshi — MoshiModel (Moshi 模型)
• mpnet — MPNetModel (MPNet 模型)
• mpt — MptModel (MPT 模型)
• mra — MraModel (MRA 模型)
• mt5 — MT5Model (MT5 模型)
• musicgen — MusicgenModel (MusicGen 模型)
• musicgen_melody — MusicgenMelodyModel (MusicGen Melody 模型)
• mvp — MvpModel (MVP 模型)
• nat — NatModel (NAT 模型)
• nemotron — NemotronModel (Nemotron 模型)
• nezha — NezhaModel (Nezha 模型)
• nllb-moe — NllbMoeModel (NLLB-MOE 模型)
• nystromformer — NystromformerModel (Nyströmformer 模型)
• olmo — OlmoModel (OLMo 模型)
• olmo2 — Olmo2Model (OLMo2 模型)
• olmoe — OlmoeModel (OLMoE 模型)
• omdet-turbo — OmDetTurboForObjectDetection (OmDet-Turbo 模型)
• oneformer — OneFormerModel (OneFormer 模型)
• open-llama — OpenLlamaModel (OpenLlama 模型)
• openai-gpt — OpenAIGPTModel (OpenAI GPT 模型)
• opt — OPTModel (OPT 模型)
• owlv2 — Owlv2Model (OWLv2 模型)
• owlvit — OwlViTModel (OWL-ViT 模型)
• patchtsmixer — PatchTSMixerModel (PatchTSMixer 模型)
• patchtst — PatchTSTModel (PatchTST 模型)
• pegasus — PegasusModel (Pegasus 模型)
• pegasus_x — PegasusXModel (PEGASUS-X 模型)
• perceiver — PerceiverModel (Perceiver 模型)
• persimmon — PersimmonModel (Persimmon 模型)
• phi — PhiModel (Phi 模型)
• phi3 — Phi3Model (Phi3 模型)
• phimoe — PhimoeModel (Phimoe 模型)
• pixtral — PixtralVisionModel (Pixtral 模型)
• plbart — PLBartModel (PLBart 模型)
• poolformer — PoolFormerModel (PoolFormer 模型)
• prophetnet — ProphetNetModel (ProphetNet 模型)
• pvt — PvtModel (PVT 模型)
• pvt_v2 — PvtV2Model (PVTv2 模型)
• qdqbert — QDQBertModel (QDQBert 模型)
• qwen2 — Qwen2Model (Qwen2 模型)
• qwen2_5_vl — Qwen2_5_VLModel (Qwen2_5_VL 模型)
• qwen2_audio_encoder — Qwen2AudioEncoder (Qwen2AudioEncoder 模型)
• qwen2_moe — Qwen2MoeModel (Qwen2MoE 模型)
• qwen2_vl — Qwen2VLModel (Qwen2VL 模型)
• recurrent_gemma — RecurrentGemmaModel (RecurrentGemma 模型)
• reformer — ReformerModel (Reformer 模型)
• regnet — RegNetModel (RegNet 模型)
• rembert — RemBertModel (RemBERT 模型)
• resnet — ResNetModel (ResNet 模型)
• retribert — RetriBertModel (RetriBERT 模型)
• roberta — RobertaModel (RoBERTa 模型)
• roberta-prelayernorm — RobertaPreLayerNormModel (RoBERTa-PreLayerNorm 模型)
• roc_bert — RoCBertModel (RoCBert 模型)
• roformer — RoFormerModel (RoFormer 模型)
• rt_detr — RTDetrModel (RT-DETR 模型)
• rt_detr_v2 — RTDetrV2Model (RT-DETRv2 模型)
• rwkv — RwkvModel (RWKV 模型)
• sam — SamModel (SAM 模型)
• seamless_m4t — SeamlessM4TModel (SeamlessM4T 模型)
• seamless_m4t_v2 — SeamlessM4Tv2Model (SeamlessM4Tv2 模型)
• segformer — SegformerModel (SegFormer 模型)
• seggpt — SegGptModel (SegGPT 模型)
• sew — SEWModel (SEW 模型)
• sew-d — SEWDModel (SEW-D 模型)
• siglip — SiglipModel (SigLIP 模型)
• siglip2 — Siglip2Model (SigLIP2 模型)
• siglip_vision_model — SiglipVisionModel (SiglipVisionModel 模型)
• smolvlm — SmolVLMModel (SmolVLM 模型)
• smolvlm_vision — SmolVLMVisionTransformer (SmolVLMVisionTransformer 模型)
• speech_to_text — Speech2TextModel (Speech2Text 模型)
• speecht5 — SpeechT5Model (SpeechT5 模型)
• splinter — SplinterModel (Splinter 模型)
• squeezebert — SqueezeBertModel (SqueezeBERT 模型)
• stablelm — StableLmModel (StableLm 模型)
• starcoder2 — Starcoder2Model (Starcoder2 模型)
• superglue — SuperGlueForKeypointMatching (SuperGlue 模型)
• swiftformer — SwiftFormerModel (SwiftFormer 模型)
• swin — SwinModel (Swin Transformer 模型)
• swin2sr — Swin2SRModel (Swin2SR 模型)
• swinv2 — Swinv2Model (Swin Transformer V2 模型)
• switch_transformers — SwitchTransformersModel (SwitchTransformers 模型)
• t5 — T5Model (T5 模型)
• table-transformer — TableTransformerModel (Table Transformer 模型)
• tapas — TapasModel (TAPAS 模型)
• textnet — TextNetModel (TextNet 模型)
• time_series_transformer — TimeSeriesTransformerModel (Time Series Transformer 模型)
• timesformer — TimesformerModel (TimeSformer 模型)
• timm_backbone — TimmBackbone (TimmBackbone 模型)
• timm_wrapper — TimmWrapperModel (TimmWrapperModel 模型)
• trajectory_transformer — TrajectoryTransformerModel (Trajectory Transformer 模型)
• transfo-xl — TransfoXLModel (Transformer-XL 模型)
• tvlt — TvltModel (TVLT 模型)
• tvp — TvpModel (TVP 模型)
• udop — UdopModel (UDOP 模型)
• umt5 — UMT5Model (UMT5 模型)
• unispeech — UniSpeechModel (UniSpeech 模型)
• unispeech-sat — UniSpeechSatModel (UniSpeechSat 模型)
• univnet — UnivNetModel (UnivNet 模型)
• van — VanModel (VAN 模型)
• videomae — VideoMAEModel (VideoMAE 模型)
• vilt — ViltModel (ViLT 模型)
• vision-text-dual-encoder — VisionTextDualEncoderModel (VisionTextDualEncoder 模型)
• visual_bert — VisualBertModel (VisualBERT 模型)
• vit — ViTModel (ViT 模型)
• vit_hybrid — ViTHybridModel (ViT Hybrid 模型)
• vit_mae — ViTMAEModel (ViTMAE 模型)
• vit_msn — ViTMSNModel (ViTMSN 模型)
• vitdet — VitDetModel (VitDet 模型)
• vits — VitsModel (VITS 模型)
• vivit — VivitModel (ViViT 模型)
• wav2vec2 — Wav2Vec2Model (Wav2Vec2 模型)
• wav2vec2-bert — Wav2Vec2BertModel (Wav2Vec2-BERT 模型)
• wav2vec2-conformer — Wav2Vec2ConformerModel (Wav2Vec2-Conformer 模型)
• wavlm — WavLMModel (WavLM 模型)
• whisper — WhisperModel (Whisper 模型)
• xclip — XCLIPModel (X-CLIP 模型)
• xglm — XGLMModel (XGLM 模型)
• xlm — XLMModel (XLM 模型)
• xlm-prophetnet — XLMProphetNetModel (XLM-ProphetNet 模型)
• xlm-roberta — XLMRobertaModel (XLM-RoBERTa 模型)
• xlm-roberta-xl — XLMRobertaXLModel (XLM-RoBERTa-XL 模型)
• xlnet — XLNetModel (XLNet 模型)
• xmod — XmodModel (X-MOD 模型)
• yolos — YolosModel (YOLOS 模型)
• yoso — YosoModel (YOSO 模型)
• zamba — ZambaModel (Zamba 模型)
• zamba2 — Zamba2Model (Zamba2 模型)

模型默认设置为评估模式，使用 model.eval() (例如，dropout 模块被禁用)。要训练模型，您应该首先使用 model.train() 将其设置回训练模式。

>>> from transformers import AutoConfig, AutoModel

>>> # Download model and configuration from huggingface.co and cache.
>>> model = AutoModel.from_pretrained("google-bert/bert-base-cased")

>>> # Update configuration during loading
>>> model = AutoModel.from_pretrained("google-bert/bert-base-cased", output_attentions=True)
>>> model.config.output_attentions
True

>>> # Loading from a TF checkpoint file instead of a PyTorch model (slower)
>>> config = AutoConfig.from_pretrained("./tf_model/bert_tf_model_config.json")
>>> model = AutoModel.from_pretrained(
...     "./tf_model/bert_tf_checkpoint.ckpt.index", from_tf=True, config=config
... )

从配置实例化库的基础模型类之一。

注意： 从其配置文件加载模型**不会**加载模型权重。它仅影响模型的配置。使用 from_pretrained() 加载模型权重。

>>> from transformers import AutoConfig, TFAutoModel

>>> # Download configuration from huggingface.co and cache.
>>> config = AutoConfig.from_pretrained("google-bert/bert-base-cased")
>>> model = TFAutoModel.from_config(config)

从预训练模型实例化库的基础模型类之一。

要实例化的模型类是根据配置对象的 model_type 属性（作为参数传递或从 pretrained_model_name_or_path 加载，如果可能）选择的，或者当它丢失时，通过回退到在 pretrained_model_name_or_path 上使用模式匹配来选择。

• albert — TFAlbertModel (ALBERT 模型)
• bart — TFBartModel (BART 模型)
• bert — TFBertModel (BERT 模型)
• blenderbot — TFBlenderbotModel (Blenderbot 模型)
• blenderbot-small — TFBlenderbotSmallModel (BlenderbotSmall 模型)
• blip — TFBlipModel (BLIP 模型)
• camembert — TFCamembertModel (CamemBERT 模型)
• clip — TFCLIPModel (CLIP 模型)
• convbert — TFConvBertModel (ConvBERT 模型)
• convnext — TFConvNextModel (ConvNeXT 模型)
• convnextv2 — TFConvNextV2Model (ConvNeXTV2 模型)
• ctrl — TFCTRLModel (CTRL 模型)
• cvt — TFCvtModel (CvT 模型)
• data2vec-vision — TFData2VecVisionModel (Data2VecVision 模型)
• deberta — TFDebertaModel (DeBERTa 模型)
• deberta-v2 — TFDebertaV2Model (DeBERTa-v2 模型)
• deit — TFDeiTModel (DeiT 模型)
• distilbert — TFDistilBertModel (DistilBERT 模型)
• dpr — TFDPRQuestionEncoder (DPR 模型)
• efficientformer — TFEfficientFormerModel (EfficientFormer 模型)
• electra — TFElectraModel (ELECTRA 模型)
• esm — TFEsmModel (ESM 模型)
• flaubert — TFFlaubertModel (FlauBERT 模型)
• funnel — TFFunnelModel 或 TFFunnelBaseModel (Funnel Transformer 模型)
• gpt-sw3 — TFGPT2Model (GPT-Sw3 模型)
• gpt2 — TFGPT2Model (OpenAI GPT-2 模型)
• gptj — TFGPTJModel (GPT-J 模型)
• groupvit — TFGroupViTModel (GroupViT 模型)
• hubert — TFHubertModel (Hubert 模型)
• idefics — TFIdeficsModel (IDEFICS 模型)
• layoutlm — TFLayoutLMModel (LayoutLM 模型)
• layoutlmv3 — TFLayoutLMv3Model (LayoutLMv3 模型)
• led — TFLEDModel (LED 模型)
• longformer — TFLongformerModel (Longformer 模型)
• lxmert — TFLxmertModel (LXMERT 模型)
• marian — TFMarianModel (Marian 模型)
• mbart — TFMBartModel (mBART 模型)
• mistral — TFMistralModel (Mistral 模型)
• mobilebert — TFMobileBertModel (MobileBERT 模型)
• mobilevit — TFMobileViTModel (MobileViT 模型)
• mpnet — TFMPNetModel (MPNet 模型)
• mt5 — TFMT5Model (MT5 模型)
• openai-gpt — TFOpenAIGPTModel (OpenAI GPT 模型)
• opt — TFOPTModel (OPT 模型)
• pegasus — TFPegasusModel (Pegasus 模型)
• regnet — TFRegNetModel (RegNet 模型)
• rembert — TFRemBertModel (RemBERT 模型)
• resnet — TFResNetModel (ResNet 模型)
• roberta — TFRobertaModel (RoBERTa 模型)
• roberta-prelayernorm — TFRobertaPreLayerNormModel (RoBERTa-PreLayerNorm 模型)
• roformer — TFRoFormerModel (RoFormer 模型)
• sam — TFSamModel (SAM 模型)
• segformer — TFSegformerModel (SegFormer 模型)
• speech_to_text — TFSpeech2TextModel (Speech2Text 模型)
• swiftformer — TFSwiftFormerModel (SwiftFormer 模型)
• swin — TFSwinModel (Swin Transformer 模型)
• t5 — TFT5Model (T5 模型)
• tapas — TFTapasModel (TAPAS 模型)
• transfo-xl — TFTransfoXLModel (Transformer-XL 模型)
• vision-text-dual-encoder — TFVisionTextDualEncoderModel (VisionTextDualEncoder 模型)
• vit — TFViTModel (ViT 模型)
• vit_mae — TFViTMAEModel (ViTMAE 模型)
• wav2vec2 — TFWav2Vec2Model (Wav2Vec2 模型)
• whisper — TFWhisperModel (Whisper 模型)
• xglm — TFXGLMModel (XGLM 模型)
• xlm — TFXLMModel (XLM 模型)
• xlm-roberta — TFXLMRobertaModel (XLM-RoBERTa 模型)
• xlnet — TFXLNetModel (XLNet 模型)

>>> from transformers import AutoConfig, TFAutoModel

>>> # Download model and configuration from huggingface.co and cache.
>>> model = TFAutoModel.from_pretrained("google-bert/bert-base-cased")

>>> # Update configuration during loading
>>> model = TFAutoModel.from_pretrained("google-bert/bert-base-cased", output_attentions=True)
>>> model.config.output_attentions
True

>>> # Loading from a PyTorch checkpoint file instead of a TensorFlow model (slower)
>>> config = AutoConfig.from_pretrained("./pt_model/bert_pt_model_config.json")
>>> model = TFAutoModel.from_pretrained(
...     "./pt_model/bert_pytorch_model.bin", from_pt=True, config=config
... )

从配置实例化库的基础模型类之一。

注意： 从其配置文件加载模型**不会**加载模型权重。它仅影响模型的配置。使用 from_pretrained() 加载模型权重。

>>> from transformers import AutoConfig, FlaxAutoModel

>>> # Download configuration from huggingface.co and cache.
>>> config = AutoConfig.from_pretrained("google-bert/bert-base-cased")
>>> model = FlaxAutoModel.from_config(config)

从预训练模型实例化库的基础模型类之一。

要实例化的模型类是根据配置对象的 model_type 属性（作为参数传递或从 pretrained_model_name_or_path 加载，如果可能）选择的，或者当它丢失时，通过回退到在 pretrained_model_name_or_path 上使用模式匹配来选择。

• albert — FlaxAlbertModel (ALBERT 模型)
• bart — FlaxBartModel (BART 模型)
• beit — FlaxBeitModel (BEiT 模型)
• bert — FlaxBertModel (BERT 模型)
• big_bird — FlaxBigBirdModel (BigBird 模型)
• blenderbot — FlaxBlenderbotModel (Blenderbot 模型)
• blenderbot-small — FlaxBlenderbotSmallModel (BlenderbotSmall 模型)
• bloom — FlaxBloomModel (BLOOM 模型)
• clip — FlaxCLIPModel (CLIP 模型)
• dinov2 — FlaxDinov2Model (DINOv2 模型)
• distilbert — FlaxDistilBertModel (DistilBERT 模型)
• electra — FlaxElectraModel (ELECTRA 模型)
• gemma — FlaxGemmaModel (Gemma 模型)
• gpt-sw3 — FlaxGPT2Model (GPT-Sw3 模型)
• gpt2 — FlaxGPT2Model (OpenAI GPT-2 模型)
• gpt_neo — FlaxGPTNeoModel (GPT Neo 模型)
• gptj — FlaxGPTJModel (GPT-J 模型)
• llama — FlaxLlamaModel (LLaMA 模型)
• longt5 — FlaxLongT5Model (LongT5 模型)
• marian — FlaxMarianModel (Marian 模型)
• mbart — FlaxMBartModel (mBART 模型)
• mistral — FlaxMistralModel (Mistral 模型)
• mt5 — FlaxMT5Model (MT5 模型)
• opt — FlaxOPTModel (OPT 模型)
• pegasus — FlaxPegasusModel (Pegasus 模型)
• regnet — FlaxRegNetModel (RegNet 模型)
• resnet — FlaxResNetModel (ResNet 模型)
• roberta — FlaxRobertaModel (RoBERTa 模型)
• roberta-prelayernorm — FlaxRobertaPreLayerNormModel (RoBERTa-PreLayerNorm 模型)
• roformer — FlaxRoFormerModel (RoFormer 模型)
• t5 — FlaxT5Model (T5 模型)
• vision-text-dual-encoder — FlaxVisionTextDualEncoderModel (VisionTextDualEncoder 模型)
• vit — FlaxViTModel (ViT 模型)
• wav2vec2 — FlaxWav2Vec2Model (Wav2Vec2 模型)
• whisper — FlaxWhisperModel (Whisper 模型)
• xglm — FlaxXGLMModel (XGLM 模型)
• xlm-roberta — FlaxXLMRobertaModel (XLM-RoBERTa 模型)

>>> from transformers import AutoConfig, FlaxAutoModel

>>> # Download model and configuration from huggingface.co and cache.
>>> model = FlaxAutoModel.from_pretrained("google-bert/bert-base-cased")

>>> # Update configuration during loading
>>> model = FlaxAutoModel.from_pretrained("google-bert/bert-base-cased", output_attentions=True)
>>> model.config.output_attentions
True

>>> # Loading from a PyTorch checkpoint file instead of a TensorFlow model (slower)
>>> config = AutoConfig.from_pretrained("./pt_model/bert_pt_model_config.json")
>>> model = FlaxAutoModel.from_pretrained(
...     "./pt_model/bert_pytorch_model.bin", from_pt=True, config=config
... )