Algorithm · OpenAI · Hard
During this live PyTorch exercise, you will work in a shared editor, actually run the code, read tracebacks, and narrate your reasoning aloud. The session contains two independent tasks. You are evaluated on correctness, the cleanliness of your debugging process, and how clearly you explain tensor shapes and gradient behavior. Before you start, it is useful to clarify with the interviewer: For Part A, whether the model is already trained and only generation must be fixed,…
Checking your access…