PaperNova

Upload a paper,get working code.

Paper → Code reads any research paper and gives you a Jupyter notebook that implements it: runnable Python for every component, explained step by step and mapped to the paper's equations.

Upload a paperBrowse classic papers
attention-is-all-you-need-implementation.ipynb
02 · Scaled dot-product attention

Implements Eq. 1 of the paper: softmax(QKᵀ/√d_k)V, with an optional mask for the decoder.

In [2]:
def attention(q, k, v, mask=None):
    scores = q @ k.transpose(-2, -1) / math.sqrt(q.size(-1))
    if mask is not None:
        scores = scores.masked_fill(mask == 0, -1e9)
    return scores.softmax(-1) @ v
7 sections9 code cellsneeds torch, numpy, matplotlib
01How it works

From PDF to a runnable notebook in minutes.

  1. 01

    Upload the paper

    Any research paper as a PDF, or pick one from the PaperNova library.

  2. 02

    We implement it

    AI reads the paper and writes working Python for each component, with explanations that point back to the paper's equations and sections.

  3. 03

    Download the notebook

    A runnable .ipynb with setup, the full implementation and a small demo that proves it works. Open it in Jupyter, VS Code or Colab.

AI-generated code: review it before relying on it. Works best with text-based PDFs.