mindformers.models¶

models init

`mindformers.models.BaseConfig`	Base Config for all models' config
`mindformers.models.BaseImageProcessor`	BaseImageProcessor for all image preprocess.
`mindformers.models.BaseModel`	The base model that contains the class method from_pretained and save_pretrained, any new model that should inherit the class.
`mindformers.models.BaseProcessor`	Base processor
`mindformers.models.Tokenizer`	Base class for all slow tokenizers.

mindformers.models.bert¶

`mindformers.models.bert.BertConfig`	BERT config class which defines the model size
`mindformers.models.bert.BertModel`	Bidirectional Encoder Representations from Transformers.
`mindformers.models.bert.BertForMultipleChoice`	Bert with dense layer for txt classification task.
`mindformers.models.bert.BertForPreTraining`	Provide bert pre-training loss through network.
`mindformers.models.bert.BertForQuestionAnswering`	Bert with dense layer for question answering task.
`mindformers.models.bert.BertForTokenClassification`	Bert with dense layer for name entity recoginition task.
`mindformers.models.bert.BertTokenizer`	Construct a BERT tokenizer.
`mindformers.models.bert.BertProcessor`	Bert processor, consists of a tokenizer (BaseTokenizer) for text input.

`mindformers.models.t5.T5Config`	T5 config class which defines the model size
`mindformers.models.t5.T5ForConditionalGeneration`	A T5 model with the loss added.
`mindformers.models.t5.T5Processor`	T5 processor, consists of a tokenizer (BaseTokenizer) for text input.
`mindformers.models.t5.T5Tokenizer`	Construct a T5 tokenizer.

`mindformers.models.clip.CLIPConfig`	Config For CLIP Model
`mindformers.models.clip.CLIPVisionConfig`	Config For CLIP Vision Module
`mindformers.models.clip.CLIPTextConfig`	Config For CLIP Text Module
`mindformers.models.clip.CLIPModel`	CLIPModel.
`mindformers.models.clip.CLIPTokenizer`	CLIP Tokenizer
`mindformers.models.clip.CLIPProcessor`	CLIP Processor, consists of a feature extractor (BaseFeatureEXtractor) for image input, and a tokenizer (BaseTokenizer) for text input.
`mindformers.models.clip.CLIPImageProcessor`	CLIPImageProcessor.

`mindformers.models.mae.ViTMAEConfig`	Config for Mae model
`mindformers.models.mae.ViTMAEForPreTraining`	Pretrain MAE Module.
`mindformers.models.mae.ViTMAEProcessor`	ViTMAEProcessor, consists of a feature extractor (BaseFeatureEXtractor) for image input.
`mindformers.models.mae.ViTMAEImageProcessor`	ViTMAEImageProcessor.

`mindformers.models.swin.SwinConfig`	Swin config class which defines the model size
`mindformers.models.swin.SwinModel`	Swin Transformer.
`mindformers.models.swin.SwinForImageClassification`	Swin Transformer Model.
`mindformers.models.swin.SwinImageProcessor`	SwinImageProcessor.
`mindformers.models.swin.SwinProcessor`	Swin processor, consists of a feature extractor (BaseFeatureEXtractor) for image input.

`mindformers.models.vit.ViTConfig`	Config for ViT model
`mindformers.models.vit.ViTModel`	Vision Transformer with support for patch or hybrid CNN input stage.
`mindformers.models.vit.ViTProcessor`	Vit processor, consists of a feature extractor (BaseFeatureEXtractor) for image input, and a tokenizer (BaseTokenizer) for text input.
`mindformers.models.vit.ViTImageProcessor`	ViTImageProcessor.
`mindformers.models.vit.ViTForImageClassification`	Vision Transformer with support for patch or hybrid CNN input stage.

`mindformers.models.gpt2.GPT2Config`	Gpt config class which defines the model size
`mindformers.models.gpt2.GPT2Model`	The backbone of GPT network
`mindformers.models.gpt2.GPT2LMHeadModel`	Provide gpt training loss or logits through network. Args: config (GPT2Config): The config of Gpt2Model.
`mindformers.models.gpt2.GPT2Tokenizer`	Construct a GPT-2 tokenizer.
`mindformers.models.gpt2.GPT2Processor`	GPT2 processor, consists of a tokenizer (BaseTokenizer) for text input.

`mindformers.models.glm.GLMConfig`	GLM config class which defines the model size
`mindformers.models.glm.GLMChatModel`	Provide glm chat capability through network. Args: config (GLMConfig): The config of GLMModel.
`mindformers.models.glm.GLMForPreTraining`	Provide glm training loss or logits through network.
`mindformers.models.glm.GLMForPreTrainingWithLora`	GLM Model for pretraining with LoRA
`mindformers.models.glm.GLMChatModelWithLora`	GLM Model for pretraining with LoRA
`mindformers.models.glm.ChatGLMTokenizer`	Construct a ChatGLM tokenizer.
`mindformers.models.glm.GLMProcessor`	GLM processor, consists of a tokenizer (BaseTokenizer) for text input.

`mindformers.models.llama.LlamaConfig`	LLaMA config class which defines the model size.
`mindformers.models.llama.LlamaModel`	Transformer decoder consisting of config.num_hidden_layers layers. Each layer is a [LlamaDecoderLayer] Args: config(LlamaConfig): the config of network.
`mindformers.models.llama.LlamaForCausalLM`	Provide llama training loss or logits through network. Args: config (LlamaConfig): The config of llama model.
`mindformers.models.llama.LlamaForCausalLMWithLora`	Llama Model for finetuning with LoRA
`mindformers.models.llama.LlamaTokenizer`	Tokenize the input string and convert them into the ids.
`mindformers.models.llama.LlamaProcessor`	Llama processor, consists of a tokenizer (BaseTokenizer) for text input.

`mindformers.models.bloom.BloomConfig`	Bloom config class which defines the model size
`mindformers.models.bloom.BloomModel`	The backbone of Bloom network
`mindformers.models.bloom.BloomLMHeadModel`	Provide bloom training loss or logits through network.
`mindformers.models.bloom.BloomTokenizer`	Tokenize the input string and convert them into the ids.
`mindformers.models.bloom.BloomProcessor`	Bloom processor, consists of a tokenizer (BaseTokenizer) for text input.