|
|
Accuracy – Multiclass
|
Easy
|
Primitives
|
|
|
|
Adapter Layer
|
Medium
|
End-to-End
|
|
|
|
ALiBi Position Bias
|
Medium
|
Research
|
|
|
|
Apply Along an Axis
|
Medium
|
Framework
|
|
|
|
AUC-ROC
|
Medium
|
Primitives
|
|
|
|
Autoencoder
|
Medium
|
End-to-End
|
|
|
|
Batched Matrix Multiply
|
Medium
|
Framework
|
|
|
|
Beam Search
|
Hard
|
End-to-End
|
|
|
|
Bidirectional RNN
|
Medium
|
Framework
|
|
|
|
Binary Classifier
|
Easy
|
End-to-End
|
|
|
|
BLEU-1gram
|
Medium
|
Primitives
|
|
|
|
BPE Encode Text
|
Hard
|
End-to-End
|
|
|
|
BPE Merge Step
|
Medium
|
Primitives
|
|
|
|
Broadcasting Addition
|
Easy
|
Framework
|
|
|
|
Causal Attention Mask
|
Easy
|
Primitives
|
|
|
|
Causal LM Forward Pass
|
Medium
|
End-to-End
|
|
|
|
Causal LM with KV Cache Generation
|
Hard
|
End-to-End
|
|
|
|
Causal Self-Attention Block
|
Medium
|
End-to-End
|
|
|
|
Classifier-Free Guidance
|
Medium
|
Primitives
|
|
|
|
Compute Gradient
|
Medium
|
Framework
|
|
|
|
Compute Jacobian
|
Hard
|
Framework
|
|
|
|
Concatenate Tensors
|
Easy
|
Framework
|
|
|
|
Confusion Matrix
|
Easy
|
Primitives
|
|
|
|
Contrastive Loss (InfoNCE)
|
Medium
|
Research
|
|
|
|
Create a Tensor from a List
|
Easy
|
Framework
|
|
|
|
Cross Attention
|
Medium
|
Research
|
|
|
|
Cross-Attention Block
|
Medium
|
End-to-End
|
|
|
|
Cumulative Sum
|
Easy
|
Framework
|
|
|
|
Custom Activation with Gradient
|
Hard
|
Framework
|
|
|
|
Data Augmentation
|
Medium
|
End-to-End
|
|
|
|
Data Collator with Padding
|
Medium
|
Primitives
|
|
|
|
DDPM Denoising Step
|
Hard
|
Primitives
|
|
|
|
DDPM Forward Noising
|
Medium
|
Primitives
|
|
|
|
DDPM Noise Schedule
|
Easy
|
Primitives
|
|
|
|
DDPM Sampler Loop
|
Hard
|
End-to-End
|
|
|
|
DDPM Training Step End-to-End
|
Medium
|
End-to-End
|
|
|
|
Depthwise Separable Convolution
|
Hard
|
Research
|
|
|
|
Distributed Training Step End-to-End
|
Hard
|
End-to-End
|
|
|
|
Efficient Attention with Masking
|
Hard
|
Framework
|
|
|
|
Einstein Summation
|
Medium
|
Framework
|
|
|
|
Element-wise Operations
|
Easy
|
Framework
|
|
|
|
Encoder-Decoder Beam Search
|
Hard
|
End-to-End
|
|
|
|
Encoder-Decoder Greedy Decode
|
Medium
|
End-to-End
|
|
|
|
Encoder-Decoder Transformer Forward Pass
|
Hard
|
End-to-End
|
|
|
|
Eval Loop with Metrics
|
Medium
|
End-to-End
|
|
|
|
Expected Calibration Error
|
Medium
|
Primitives
|
|
|
|
Exponential Moving Average
|
Medium
|
Primitives
|
|
|
|
Feature Normalization Pipeline
|
Easy
|
End-to-End
|
|
|
|
Flash Attention Score Computation
|
Hard
|
Research
|
|
|
|
Focal Loss
|
Medium
|
Research
|
|