+ jDNRt^RIt^RIHt^RIHt^RIHt^RI H t ^RI H t H t ^RIHt^RIHt^RIHtHtHt^R IHt^R IHt^R IHtHtHtHt^R IH t H!t!H"t"^R I#H$t$H%t%H&t&H't'H(t(H)t)H*t*H+t+H,t,H-t-^RI.H/t/H0t0H1t1H2t2H3t3]!RR7]!RR]!444t4]!RR7]!RR]"444t5]!RR7]!RR] 444t6!RR](4t7!RR]/4t8!RR]4t9!RR]34t:!RR ]4t;!R!R"]04t<!R#R$]14t=!R%R&]4t>!R'R(]&4t?!R)R*]24t@!R+R,]%4tA!R-R.]$4tB!R/R0]'4tC!R1R2]&4tD!R3R4]*4tE!R5R6])4tF!R7R8]+4tG]!R9R:7!R;R<]F44tH!R=R>]F4tI!R?R@]]F4tJ.RAOtKR#)BzPyTorch AltCLIP model.N)strict)initialization)create_bidirectional_mask)BaseModelOutputWithPooling'BaseModelOutputWithPoolingAndProjection)PreTrainedModel)Unpack)TransformersKwargsauto_docstringcan_return_tuple)merge_with_config_defaults)capture_outputs)ChineseCLIPModelChineseCLIPTextAttentionChineseCLIPTextLayerChineseCLIPTextSelfAttention) CLIPConfigCLIPTextConfigCLIPVisionConfig) CLIPMLP CLIPAttention CLIPEncoderCLIPEncoderLayer CLIPOutputCLIPPreTrainedModelCLIPVisionEmbeddingsCLIPVisionModel_get_vector_normimage_text_contrastive_loss)RobertaEmbeddingsRobertaIntermediate RobertaOutput RobertaPoolerRobertaSelfOutputz BAAI/AltCLIP) checkpointca]tRt^:toRtRtRt^t^tRt Rt Rt ^t Rt ^tRt^t^t^tR t]!4t]!4tV3R ltR tVtR #) AltCLIPTextConfiga" project_dim (`int`, *optional*, defaults to 768): The dimensions of the teacher model before the mapping layer. Examples: ```python >>> from transformers import AltCLIPTextModel, AltCLIPTextConfig >>> # Initializing a AltCLIPTextConfig with BAAI/AltCLIP style configuration >>> configuration = AltCLIPTextConfig() >>> # Initializing a AltCLIPTextModel (with random weights) from the BAAI/AltCLIP style configuration >>> model = AltCLIPTextModel(configuration) >>> # Accessing the model configuration >>> configuration = model.config ```iiigelug?ig{Gz?c<V^8dQh/S[;R&S[;R&S[;R&S[;R&S[;R&S[;R&S[S[,;R&S[S[,;R&S[;R &S[;R &S[;R &S[R ,;R &S[R ,;R&S[R ,;R&S[;R&#) vocab_size hidden_sizenum_hidden_layersnum_attention_headsintermediate_size hidden_acthidden_dropout_probattention_probs_dropout_probmax_position_embeddingstype_vocab_sizeinitializer_factorN pad_token_id bos_token_id eos_token_id project_dim)intstrfloat)format __classdict__s"l/home/luca/screen-checking/.venv/lib/python3.14/site-packages/transformers/models/altclip/modular_altclip.py __annotate__AltCLIPTextConfig.__annotate__:s,-./012!34!5678u*9:#&+1;<!&=>?@$AB* CD* EF* GHIN)__name__ __module__ __qualname____firstlineno____doc__r+r,r-r.r/r0r1r2r3r4r5r6r7r8r9AttributeErrorprojection_dimattention_dropout__annotate_func____static_attributes____classdictcell__r>s@r?r&r&:sw &JK!!J'*01 #&O $ L L LK#%N&(MrBr&c]tRt^ctRtRtR#)AltCLIPVisionConfiga Example: ```python >>> from transformers import AltCLIPVisionConfig, AltCLIPVisionModel >>> # Initializing a AltCLIPVisionConfig with BAAI/AltCLIP style configuration >>> configuration = AltCLIPVisionConfig() >>> # Initializing a AltCLIPVisionModel (with random weights) from the BAAI/AltCLIP style configuration >>> model = AltCLIPVisionModel(configuration) >>> # Accessing the model configuration >>> configuration = model.config ```rCN)rDrErFrGrHrMrCrBr?rQrQcs rBrQc4a]tRt^wtoRtRtV3RltRtVtR#) AltCLIPConfiga Example: ```python >>> from transformers import AltCLIPConfig, AltCLIPModel >>> # Initializing a AltCLIPConfig with BAAI/AltCLIP style configuration >>> configuration = AltCLIPConfig() >>> # Initializing a AltCLIPModel (with random weights) from the BAAI/AltCLIP style configuration >>> model = AltCLIPModel(configuration) >>> # Accessing the model configuration >>> configuration = model.config >>> # We can also initialize a AltCLIPConfig from a AltCLIPTextConfig and a AltCLIPVisionConfig >>> # Initializing a AltCLIPText and AltCLIPVision configuration >>> config_text = AltCLIPTextConfig() >>> config_vision = AltCLIPVisionConfig() >>> config = AltCLIPConfig(text_config=config_text, vision_config=config_vision) ```r(c&<V^8dQh/S[;R&#)r*rJ)r:)r=r>s"r?r@AltCLIPConfig.__annotate__ws67rBrCN) rDrErFrGrHrJrLrMrNrOs@r?rSrSws 0N7rBrSc]tRt^tRtR#) AltCLIPOutputrCNrDrErFrGrMrCrBr?rWrWrBrWc]tRt^tRtR#)AltRobertaEmbeddingsrCNrXrCrBr?r[r[rYrBr[c2aa]tRt^toV3RltRtVtV;t#)AltRobertaSelfAttentionc4<\SV`V4RVnR#)FN)super__init__ is_causalselfconfig __class__s&&r?r` AltRobertaSelfAttention.__init__s  rB)rarDrErFrGr`rMrN __classcell__rer>s@@r?r]r]srBr]c]tRt^tRtR#)AltRobertaSelfOutputrCNrXrCrBr?rkrkrYrBrkc2aa]tRt^toV3RltRtVtV;t#)AltRobertaAttentioncd<\SV`4\V4Vn\ V4VnR#N)r_r`r]rcrkoutputrbs&&r?r`AltRobertaAttention.__init__s& +F3 *62 rB)rprcrgris@@r?rmrms33rBrmc]tRt^tRtR#)AltRobertaIntermediaterCNrXrCrBr?rsrsrYrBrsc]tRt^tRtR#)AltRobertaOutputrCNrXrCrBr?rururYrBruc2aa]tRt^toV3RltRtVtV;t#)AltRobertaLayerc<\SV`4\V4Vn\ V4Vn\ V4VnR#ro)r_r`rm attentionrs intermediaterurprbs&&r?r`AltRobertaLayer.__init__s3 ,V426:&v. rB)ryrzrprgris@@r?rwrws//rBrwcBaa]tRt^toRtV3RlV3RlltRtVtV;t#)AltRobertaEncoderz Transformer encoder consisting of `config.num_hidden_layers` self attention layers. Each layer is a [`AltRobertaEncoderLayer`]. Args: config: AltCLIPTextConfig c <V^8dQhRS[/#r*rdr&)r=r>s"r?r@AltRobertaEncoder.__annotate__shh0hrBc<\SV`V4\P!\ VP 4Uu.uFp\ V4NK up4VnR#uupiro)r_r`nn ModuleListranger-rwlayers)rcrd_res&& r?r`AltRobertaEncoder.__init__sD  mmeFLdLdFe$fFe_V%aa]tRt^toV3RlV3RlltRtVtV;t#)AltCLIPEncoderLayerc <V^8dQhRS[/#r)rQ)r=r>s"r?r@ AltCLIPEncoderLayer.__annotate__s!!2!rBc&<\SV`V4R#ro)r_r`rbs&&r?r`AltCLIPEncoderLayer.__init__s  rBrCrgris@@r?rrs!!!rBrc]tRt^tRtR#)AltCLIPEncoderrCNrXrCrBr?rrrYrBrc]tRt^tRtR#)AltCLIPVisionEmbeddingsrCNrXrCrBr?rrrYrBrcba]tRt^to.ROtR]R]/t]P!4R4t Rt Vt R#)AltCLIPPreTrainedModel hidden_states attentionsc H \P!W4VPPp\ V\ 4'Ed\ P!VPRVPR,V,R7\ P!VPPVPPV,R7\ P!VPPVPPV,R7\ P!VP\ P"!VP$4P'R44R#\ V\(4'EdVPR,^VPP*,R,,V,pVPR,V,p\ P!VP,PVR7\ P!VP.PVR7\ P!VP0PVR7\ P!VP2PVR7R#\ V\44'dVPP6R,^VPP*,R,,V,p^VPP6,R,V,p\ P!VP8PVR7\ P!VP:PVR7R#\ V\<4'd\ P!VP>PVP@R,V,R7\ P!VPBPVPDR,V,R7R#\ V\F4'd\ P!VP\ P"!VPPHR,4P'R44\ PJ!VPL4R#R#)zInitialize the weightsg)meanstd)rNg)r)'r _init_weightsrdr5 isinstancerinitnormal_class_embedding embed_dimpatch_embeddingweightinitializer_rangeposition_embeddingcopy_ position_idstorcharange num_positionsexpandrr-q_projk_projv_projout_projrr,fc1fc2 AltCLIPModeltext_projectiontext_embed_dimvisual_projectionvision_embed_dimr[shapezeros_token_type_ids)rcmodulefactor in_proj_std out_proj_stdfc_stds&& r?r$AltCLIPPreTrainedModel._init_weightss %%d3// f5 6 6 LL//cv?O?OQU?UX^?^ _ LL//66FMM<[<[^d>> import httpx >>> from io import BytesIO >>> from PIL import Image >>> from transformers import AutoProcessor, AltCLIPVisionModel >>> model = AltCLIPVisionModel.from_pretrained("BAAI/AltCLIP") >>> processor = AutoProcessor.from_pretrained("BAAI/AltCLIP") >>> url = "http://images.cocodataset.org/val2017/000000039769.jpg" >>> with httpx.stream("GET", url) as response: ... image = Image.open(BytesIO(response.read())) >>> inputs = processor(images=image, return_tensors="pt") >>> outputs = model(**inputs) >>> last_hidden_state = outputs.last_hidden_state >>> pooled_output = outputs.pooler_output # pooled CLS states ```rC)r_forwardrc super_kwargsres&,r?rAltCLIPVisionModel.forwards.w...rBrC)rDrErFrGrrMrNrhris@@r?rrs//rBraE The model behaves as an encoder following the architecture described in *Attention is all you need*_ by Ashish Vaswani, Noam Shazeer, Niki Parmar, Jakob Uszkoreit, Llion Jones, Aidan N. Gomez, Lukasz Kaiser and Illia Polosukhin. .. _*Attention is all you need*: https://huggingface.co/papers/1706.03762 ) custom_introcaa]tRtRtoR tRtR]R]/tR V3Rllt ] ] ] R V3RlRll444t V3RltR tVtV;t#) AltRobertaModeli+word_embeddingsrrc <\SV`V4\V4Vn\ V4VnV'd \ V4MRVnVP4R#)z^ add_pooling_layer (bool, *optional*, defaults to `True`): Whether to add a pooling layer N) r_r`r[ embeddingsr}encoderrpooler post_init)rcrdadd_pooling_layerres&&&r?r`AltRobertaModel.__init__>sE  .v6(0 2C&v.  rBc<V^8dQhRS[PR,RS[PR,RS[PR,RS[PR,RS[PR,RS[S[,RS[S[,/# r* input_idsNattention_maskrr inputs_embedskwargsreturn)rTensorrr tupler)r=r>s"r?r@AltRobertaModel.__annotate__Ms3 3 <<$&3  t+3  t+ 3 llT) 3 ||d* 3 +,3  + +3 rBc 0VRJVRJ, 'd \R4hVPVVVVR7p\VPVVR7pVP!V3RV/VBpV^,pVP eVP V4MRp \ VV R7#)a Examples: ```python >>> from transformers import AutoTokenizer, AltRobertaModel >>> model = AltRobertaModel.from_pretrained("openai/alt_roberta-vit-base-patch32") >>> tokenizer = AutoTokenizer.from_pretrained("openai/alt_roberta-vit-base-patch32") >>> inputs = tokenizer(["a photo of a cat", "a photo of a dog"], padding=True, return_tensors="pt") >>> outputs = model(**inputs) >>> last_hidden_state = outputs.last_hidden_state >>> pooled_output = outputs.pooler_output # pooled (EOS token) states ```Nz:You must specify exactly one of input_ids or inputs_embeds)rrrr)rdrrr)last_hidden_state pooler_output) ValueErrorrrrdrrr) rcrrrrrrencoder_outputssequence_output pooled_outputs &&&&&&, r?rAltRobertaModel.forwardJs6  -t"; < <YZ Z%)' ( 3;;')  ,,  )   *!,8< 8O O4UY )-'  rBc&<V^8dQh/S[;R&#rr)r=r>s"r?r@r+s rB)rrrtext)TNNNNN)rDrErFrGinput_modalities_input_embed_layerrwr]rr`r r r rrLrMrNrhris@@r?rr+s\!*-  3 3  3 ErBrctaa]tRtRtoR tRtRtV3Rlt]] R V3RlRll44t V3Rlt Rt Vt V;t#) AltCLIPTextModelirrobertac,<\SV`V4\VRR7Vn\P !VP VP4Vn\P!VP VPR7Vn VP4R#)F)r)epsN) r_r`rrrLinearr,r9transformation LayerNormlayer_norm_epspre_LNrrbs&&r?r`AltCLIPTextModel.__init__se  &vG  ii(:(:Fs"r?r@AltCLIPTextModel.__annotate__s3 3 <<$&3  t+3  t+ 3 llT) 3 ||d* 3 +,3  8 83 rBc VP!RRVRVRVRVRV/VBpV^,pVPV4pVPV4p V R,p \V V VPVP R7#) a Examples: ```python >>> from transformers import AutoProcessor, AltCLIPTextModel >>> model = AltCLIPTextModel.from_pretrained("BAAI/AltCLIP") >>> processor = AutoProcessor.from_pretrained("BAAI/AltCLIP") >>> texts = ["it's a cat", "it's a dog"] >>> inputs = processor(text=texts, padding=True, return_tensors="pt") >>> outputs = model(**inputs) >>> last_hidden_state = outputs.last_hidden_state >>> pooled_output = outputs.pooler_output # pooled CLS states ```rrrrr)rrrrrC):NNN)rrrrrr) rcrrrrrroutputsrprojection_staters &&&&&&, r?rAltCLIPTextModel.forwards:,,  ) * &  (    "!*++o6 ..?(. 6.'!//))   rBc&<V^8dQh/S[;R&#rr)r=r>s"r?r@rs rB)rrrrr)rDrErFrGrrbase_model_prefixr`r r rrLrMrNrhris@@r?rrsD *!3 3 3 rBrcxaa]tRtRtoV3RlV3RlltV3RltV3RltR V3RlRlltV3RltR t Vt V;t #) ric <V^8dQhRS[/#rrS)r=r>s"r?r@AltCLIPModel.__annotate__sWW}WrBc<\SV`V4VPpVPVn\ P VPP4Vn\P VPP4Vn R#ro) r_r` text_configr9rr _from_configrd text_modelr vision_config vision_model)rcrdr res&& r?r`AltCLIPModel.__init__sb  (( )55*77 8O8OP.;;DKK>> import torch >>> from transformers import AutoProcessor, AltCLIPModel >>> model = AltCLIPModel.from_pretrained("BAAI/AltCLIP") >>> processor = AutoProcessor.from_pretrained("BAAI/AltCLIP") >>> inputs = processor(text=["a photo of a cat", "a photo of a dog"], padding=True, return_tensors="pt") >>> with torch.inference_mode(): ... text_features = model.get_text_features(**inputs) ```rC)r_get_text_featuresrs&,r?rAltCLIPModel.get_text_featuressw(8<88rBc $<\SV`!R/VB#)a Examples: ```python >>> import torch >>> from transformers import AutoProcessor, AltCLIPModel >>> from transformers.image_utils import load_image >>> model = AltCLIPModel.from_pretrained("BAAI/AltCLIP") >>> processor = AutoProcessor.from_pretrained("BAAI/AltCLIP") >>> url = "http://images.cocodataset.org/val2017/000000039769.jpg" >>> image = load_image(url) >>> inputs = processor(images=image, return_tensors="pt") >>> with torch.inference_mode(): ... image_features = model.get_image_features(**inputs) ```rC)r_get_image_featuresrs&,r?rAltCLIPModel.get_image_featuress&w)9L99rBc(<V^8dQhRS[PR,RS[PR,RS[PR,RS[PR,RS[PR,RS[R,RS[R S[S[,R S[S[,/ #) r*rN pixel_valuesrrr return_lossinterpolate_pos_encodingrr) r LongTensor FloatTensorrboolrr rrW)r=r>s"r?r@r sL L ##d*L ''$.L  t+ L  t+ L &&- L D[L #'L +,L   L rBc lVP!R RVRV/VBp VP!R RVRVRVRV/VBp V ^,p VPV 4p V ^,p VPV 4p V \ V 4, p V \ V 4, p \ P !WP4PV P44p WPP4PV P4,p V P4pRpV'd \V 4p\VVV V V V V R7#) u return_loss (`bool`, *optional*): Whether or not to return the contrastive loss. Examples: ```python >>> import torch >>> from transformers import AutoProcessor, AltCLIPModel >>> from transformers.image_utils import load_image >>> model = AltCLIPModel.from_pretrained("OFA-Sys/chinese-clip-vit-base-patch16") >>> processor = AutoProcessor.from_pretrained("OFA-Sys/chinese-clip-vit-base-patch16") >>> url = "https://clip-cn-beijing.oss-cn-beijing.aliyuncs.com/pokemon.jpeg" >>> image = load_image(url) >>> inputs = processor(text=["杰尼龟", "妙蛙种子", "小火龙", "皮卡丘"], images=image, return_tensors="pt", padding=True) >>> with torch.inference_mode(): ... outputs = model(**inputs) >>> logits_per_image = outputs.logits_per_image # this is the image-text similarity score >>> probs = logits_per_image.softmax(dim=1) # we can take the softmax to get the label probabilities ```rrrrrrN)losslogits_per_imagelogits_per_text text_embeds image_embedstext_model_outputvision_model_outputrC)rrrrrrmatmulttodevice logit_scaleexprrW)rcrrrrrrrrvision_outputs text_outputsr$r#r"r!r s&&&&&&&&, r?rAltCLIPModel.forwardsWF** % %=     ) * &    &a( --l; "1o **;7 $&6|&DD !$4[$AA  ,,{NN4D4G4G HZHZ4[\),<,<,@,@,B,E,EkFXFX,YY*,,. .?D-+#%* .  rBc&<V^8dQh/S[;R&#rr )r=r>s"r?r@r s rB)rrr)NNNNNNF) rDrErFrGr`rrrrLrMrNrhris@@r?rrs/WW9":*L L arBr)rrrrr&rQrS)LrHrtorch.nnrhuggingface_hub.dataclassesrrr masking_utilsrmodeling_outputsrrmodeling_utilsrprocessing_utilsrutilsr r r utils.genericr utils.output_capturingr "chinese_clip.modeling_chinese_cliprrrrclip.configuration_cliprrrclip.modeling_cliprrrrrrrrrrroberta.modeling_robertarr r!r"r#r&rQrSrWr[r]rkrmrsrurwr}rrrrrrrrrrr__all__rCrBr?r@s .&6.&II75 SR   >*$)$)+$)N>* * + $>*J+8 J  , :  , 323 0  } /*/ h h }  }   !*!  [  2 (/0(/V//6L ,L L ^B -B J| #%;| ~ rB