^j dZddlmZddlmZddlmZmZeje Z edeGdd eZ edeGd d eZ edeGd d eZ gdZy)zPix2Struct model configuration)strict)PreTrainedConfig)auto_docstringloggingzgoogle/pix2struct-base) checkpointcleZdZUdZdZdgZddddddddZdZee d <d Z ee d<d Z ee d <d Z ee d<dZ ee d<dZee d<dZee d<dZee d<dZeeze d<dZee d<dZee d<dZee d<dZee d<dZee d<dZed ze d!<d"Zeeezd ze d#<d Zed ze d$<dZee d%<d&Zee d'<dZ ee d(<y ))Pix2StructTextConfiga relative_attention_num_buckets (`int`, *optional*, defaults to 32): The number of buckets to use for each attention layer. relative_attention_max_distance (`int`, *optional*, defaults to 128): The maximum distance of the longer sequences for the bucket separation. dense_act_fn (`Union[Callable, str]`, *optional*, defaults to `"gelu_new"`): The non-linear activation function (function or string). Example: ```python >>> from transformers import Pix2StructTextConfig, Pix2StructTextModel >>> # Initializing a Pix2StructTextConfig with google/pix2struct-base style configuration >>> configuration = Pix2StructTextConfig() >>> # Initializing a Pix2StructTextModel (with random weights) from the google/pix2struct-base style configuration >>> model = Pix2StructTextModel(configuration) >>> # Accessing the model configuration >>> configuration = model.config ```pix2struct_text_modelpast_key_values hidden_size num_heads num_layers)r num_attention_headsnum_hidden_layersdecoder_attention_headsencoder_attention_headsencoder_layersdecoder_layersiD vocab_size@d_kvd_ff relative_attention_num_bucketsrelative_attention_max_distanceg? dropout_rateư>layer_norm_epsilon?initializer_factorgelu_new dense_act_fnrdecoder_start_token_idF use_cacheN pad_token_id eos_token_id bos_token_idtie_word_embeddingsT is_decoderadd_cross_attention)!__name__ __module__ __qualname____doc__ model_typekeys_to_ignore_at_inference attribute_maprint__annotations__r rrrrrr r!floatr#r%r'strr(r)boolr*r,listr-r.r/r0/var/www/ramen.bs-engineer-server.com/venv/lib/python3.12/site-packages/transformers/models/pix2struct/configuration_pix2struct.pyr r s .)J#4"5$*)#.#.&&MJKD#ND#JIs*,"C,+.#S. #L%#+# $$ ##"L#""#C#It L#* +,L#S /D(,#L#*# %%J %%r?r ceZdZUdZdZdZeed<dZeed<dZ eed<dZ eed <d Z eed <d Z eed <d Z eed<dZeed<dZeezed<dZeezed<dZeed<dZeed<dZeed<dZeed<dZeed<y)Pix2StructVisionConfiga patch_embed_hidden_size (`int`, *optional*, defaults to 768): Dimensionality of the input patch_embedding layer in the Transformer encoder. d_ff (`int`, *optional*, defaults to 2048): Dimensionality of the "intermediate" (i.e., feed-forward) layer in the Transformer encoder. d_kv (`int`, *optional*, defaults to 64): Dimensionality of the key, query, value projections per attention head. The non-linear activation function (function or string) in the encoder and pooler. If string, `"gelu"`, `"relu"`, `"selu"` and `"gelu_new"` `"gelu"` are supported. dense_act_fn (`Union[Callable, str]`, *optional*, defaults to `"gelu_new"`): The non-linear activation function (function or string). seq_len (`int`, *optional*, defaults to 4096): Maximum sequence length (here number of patches) supported by the model. relative_attention_num_buckets (`int`, *optional*, defaults to 32): The number of buckets to use for each attention layer. relative_attention_max_distance (`int`, *optional*, defaults to 128): The maximum distance (in tokens) to use for each attention layer. Example: ```python >>> from transformers import Pix2StructVisionConfig, Pix2StructVisionModel >>> # Initializing a Pix2StructVisionConfig with google/pix2struct-base style configuration >>> configuration = Pix2StructVisionConfig() >>> # Initializing a Pix2StructVisionModel (with random weights) from the google/pix2struct-base style configuration >>> model = Pix2StructVisionModel(configuration) >>> # Accessing the model configuration >>> configuration = model.config ```pix2struct_vision_modelrr patch_embed_hidden_sizerrrrrrrr&r'r"layer_norm_epsgr!attention_dropoutg|=initializer_ranger$r%iseq_lenrrrr N)r1r2r3r4r5r r8r9rDrrrrr'r;rEr:r!rFrGr%rHrr r>r?r@rBrBUs B+JK#&S&D#D#Ns!!"L#" NE #L%#+#%(us{($u$ ##GS*,"C,+.#S.r?rBceZdZUdZdZeedZdZe e zdze d<dZ e e zdze d<dZ ee d<d Zee d <d Zee d <d Zee d <dZee d<fdZxZS)Pix2StructConfiga is_vqa (`bool`, *optional*, defaults to `False`): Whether the model has been fine-tuned for VQA or not. Example: ```python >>> from transformers import Pix2StructConfig, Pix2StructForConditionalGeneration >>> # Initializing a Pix2StructConfig with google/pix2struct-base style configuration >>> configuration = Pix2StructConfig() >>> # Initializing a Pix2StructForConditionalGeneration (with random weights) from the google/pix2struct-base style configuration >>> model = Pix2StructForConditionalGeneration(configuration) >>> # Accessing the model configuration >>> configuration = model.config >>> # We can also initialize a Pix2StructConfig from a Pix2StructTextConfig and a Pix2StructVisionConfig >>> # Initializing a Pix2Struct text and Pix2Struct vision configuration >>> config_text = Pix2StructTextConfig() >>> config_vision = Pix2StructVisionConfig() >>> config = Pix2StructConfig(text_config=config_text, vision_config=config_vision) ``` pix2struct) text_config vision_configNrLrMr$r%g{Gz?rGFis_vqar.Tis_encoder_decoderc X|j)rLr rOr.loggerinfo isinstancedictrMrBr(r*r,rGsuper __post_init__)selfkwargs __class__s r@rVzPix2StructConfig.__post_init__sO    #3#'#:#:$($<$< D  KKo p (($ /595L5LD  1 26:6N6ND  2 33Gd6F6FGD     %!7!9D  KKs t **D 1!7!M$:L:L!MD &*&6&6&M&M# ,,99 ,,99-1-C-C*/3/E/E, ''r?)r1r2r3r4r5r rB sub_configsrLrTrr9rMr%r:rGrNr<r.rOrV __classcell__)rYs@r@rJrJs 6J"6I_`K26K((4/648M4**T18 ###u#FD %%##((r?rJ)rJr rBN)r4huggingface_hub.dataclassesrconfiguration_utilsrutilsrr get_loggerr1rQr rBrJ__all__r>r?r@ras%.3,   H %347&+7&57&t342/-2/52/j34@('@(5@(F Qr?