mmlearn.modules package

Reusable components for building tasks.

Subpackages

• mmlearn.modules.encoders package
  • Submodules
    • mmlearn.modules.encoders.clip module
      • HFCLIPTextEncoder
      • HFCLIPTextEncoderWithProjection
      • HFCLIPVisionEncoder
      • HFCLIPVisionEncoderWithProjection
    • mmlearn.modules.encoders.text module
      • HFTextEncoder
    • mmlearn.modules.encoders.vision module
      • TimmViT
      • VisionTransformer
      • VisionTransformerPredictor
      • vit_base()
      • vit_giant()
      • vit_huge()
      • vit_large()
      • vit_predictor()
      • vit_small()
      • vit_tiny()
• mmlearn.modules.layers package
  • Submodules
    • mmlearn.modules.layers.attention module
      • Attention
    • mmlearn.modules.layers.embedding module
      • ConvEmbed
      • PatchEmbed
      • get_1d_sincos_pos_embed()
      • get_1d_sincos_pos_embed_from_grid()
      • get_2d_sincos_pos_embed()
      • get_2d_sincos_pos_embed_from_grid()
    • mmlearn.modules.layers.logit_scaling module
      • LearnableLogitScaling
    • mmlearn.modules.layers.mlp module
      • MLP
    • mmlearn.modules.layers.normalization module
      • L2Norm
    • mmlearn.modules.layers.patch_dropout module
      • PatchDropout
    • mmlearn.modules.layers.transformer_block module
      • Block
      • DropPath
      • drop_path()
• mmlearn.modules.losses package
  • Submodules
    • mmlearn.modules.losses.contrastive module
      • ContrastiveLoss
    • mmlearn.modules.losses.data2vec module
      • Data2VecLoss
• mmlearn.modules.lr_schedulers package
  • Submodules
    • mmlearn.modules.lr_schedulers.linear_warmup_cosine_lr module
      • linear_warmup_cosine_annealing_lr()
• mmlearn.modules.metrics package
  • Submodules
    • mmlearn.modules.metrics.retrieval_recall module
      • RetrievalRecallAtK

Submodules

• mmlearn.modules.ema module
  • ExponentialMovingAverage
    • ExponentialMovingAverage.deepcopy_model()
    • ExponentialMovingAverage.get_annealed_rate()
    • ExponentialMovingAverage.restore()
    • ExponentialMovingAverage.state_dict()
    • ExponentialMovingAverage.step()