DeepForgeby svx

0/5730 solved

← All interview tracks

MLE · research engineersTransformers · RLHF · safety

Anthropic · Machine Learning Engineer

Attention internals, transformer training dynamics, and the RLHF stack — from softmax to DPO, with safety-flavoured objectives at the end.

3 phases · 25 problems · 25 unique

Start timed mock5 problems · 4 min each

Your progress

0/25 · 0%

Continue track · ReLU

1Warm-up

0/8

ReLU
dl-001Deep Learning
Easy
Sigmoid (Vector)
dl-002Deep Learning
Easy
Numerically Stable Softmax
dl-003Deep Learning
Medium
Causal Mask
dl-015Deep Learning
Easy
Sinusoidal Positional Encoding
dl-033Deep Learning
Medium
Scaled Attention Scores
dl-034Deep Learning
Medium
Attention Softmax Weights
dl-035Deep Learning
Medium
Perplexity From Logits
dl-041Deep Learning
Medium

2Core

0/9

Attention Weighted Sum
dl-036Deep Learning
Medium
Multi-Head Split
dl-037Deep Learning
Medium
Multi-Head Merge
dl-038Deep Learning
Medium
LayerNorm Forward
dl-027Deep Learning
Medium
Transformer Decoder Masked Self-Attention
dl-116Deep Learning
Medium
InfoNCE Loss
dl-130Deep Learning
Medium
Reward Model Loss
dl-170Deep Learning
Medium
Token-Level KL
dl-171Deep Learning
Medium
Sparse Autoencoder Loss
dl-175Deep Learning
Medium

3Hard

0/8

Softmax + Cross-Entropy Backward
dl-043Deep Learning
Hard
Label Smoothing Cross-Entropy
dl-048Deep Learning
Hard
Transformer Encoder Block Forward
dl-087Deep Learning
Hard
Spectral Norm Power Iteration
dl-089Deep Learning
Hard
Transformer Encoder Block Parameter Count
dl-091Deep Learning
Hard
Multi-Head Attention Parameter Count with Bias
dl-092Deep Learning
Hard
DPO Loss
dl-180Deep Learning
Hard
GRPO Advantage
dl-181Deep Learning
Hard