hezar.models.image2text.vit_roberta.vit_roberta_image2text_config module

class hezar.models.image2text.vit_roberta.vit_roberta_image2text_config.DecoderConfig(is_decoder: bool = True, add_cross_attention: bool = True, attention_probs_dropout_prob: float = 0.1, bos_token_id: int = 0, eos_token_id: int = 2, classifier_dropout: float = None, gradient_checkpointing: bool = False, hidden_act: str = 'gelu', hidden_dropout_prob: float = 0.1, hidden_size: int = 768, initializer_range: int = 0.02, intermediate_size: int = 3072, layer_norm_eps: float = 1e-12, max_position_embeddings: int = 514, num_attention_heads: int = 12, num_hidden_layers: int = 12, pad_token_id: int = 2, position_embedding_type: str = 'absolute', type_vocab_size: int = 1, use_cache: bool = True, vocab_size: int = 42000)

Bases: ModelConfig

add_cross_attention: bool = True

attention_probs_dropout_prob: float = 0.1

bos_token_id: int = 0

classifier_dropout: float = None

eos_token_id: int = 2

gradient_checkpointing: bool = False

hidden_act: str = 'gelu'

hidden_dropout_prob: float = 0.1

hidden_size: int = 768

initializer_range: int = 0.02

intermediate_size: int = 3072

is_decoder: bool = True

layer_norm_eps: float = 1e-12

max_position_embeddings: int = 514

num_attention_heads: int = 12

num_hidden_layers: int = 12

pad_token_id: int = 2

position_embedding_type: str = 'absolute'

type_vocab_size: int = 1

use_cache: bool = True

vocab_size: int = 42000

class hezar.models.image2text.vit_roberta.vit_roberta_image2text_config.EncoderConfig(hidden_size: int = 768, num_hidden_layers: int = 12, num_attention_heads: int = 12, intermediate_size: int = 3072, hidden_act: str = 'gelu', hidden_dropout_prob: float = 0.0, attention_probs_dropout_prob: float = 0.0, initializer_range: float = 0.02, layer_norm_eps: float = 1e-12, image_size: int = 224, patch_size: int = 16, num_channels: int = 3, qkv_bias: bool = True, encoder_stride: int = 16)

Bases: ModelConfig

attention_probs_dropout_prob: float = 0.0

encoder_stride: int = 16

hidden_act: str = 'gelu'

hidden_dropout_prob: float = 0.0

hidden_size: int = 768

image_size: int = 224

initializer_range: float = 0.02

intermediate_size: int = 3072

layer_norm_eps: float = 1e-12

num_attention_heads: int = 12

num_channels: int = 3

num_hidden_layers: int = 12

patch_size: int = 16

qkv_bias: bool = True

class hezar.models.image2text.vit_roberta.vit_roberta_image2text_config.GenerationConfig(bos_token_id: int = 0, decoder_start_token_id: int = 0, return_dict_in_generate: bool = False, early_stopping: bool = True, eos_token_id: int = 2, length_penalty: float = 2.0, max_length: int = 64, no_repeat_ngram_size: int = 3, num_beams: int = 4, pad_token_id: int = 2)

Bases: object

bos_token_id: int = 0

decoder_start_token_id: int = 0

early_stopping: bool = True

eos_token_id: int = 2

length_penalty: float = 2.0

max_length: int = 64

no_repeat_ngram_size: int = 3

num_beams: int = 4

pad_token_id: int = 2

return_dict_in_generate: bool = False

class hezar.models.image2text.vit_roberta.vit_roberta_image2text_config.ViTRobertaImage2TextConfig(encoder: hezar.models.image2text.vit_roberta.vit_roberta_image2text_config.EncoderConfig = <factory>, decoder: hezar.models.image2text.vit_roberta.vit_roberta_image2text_config.DecoderConfig = <factory>, generation: hezar.models.image2text.vit_roberta.vit_roberta_image2text_config.GenerationConfig = <factory>)

Bases: ModelConfig

decoder: DecoderConfig

encoder: EncoderConfig

generation: GenerationConfig

name: str = 'vit_roberta_image2text'