# paper2code
> **arxiv URL in โ citation-anchored implementation out**
```
โโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโ โโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโ
โ โ โ {paper_slug}/ โ
โ /paper2code โ โ โโโ README.md โ
โ https://arxiv.org/abs/ โ โโโโถ โ โโโ REPRODUCTION_NOTES.md โ
โ 1706.03762 โ โ โโโ requirements.txt โ
โ โ โ โโโ src/ โ
โ โ โ โ โโโ model.py # ยง3.2 cited โ
โ โ โ โ โโโ loss.py # ยง3.4 cited โ
โ โ โ โ โโโ train.py # ยง4.1 cited โ
โ โ โ โ โโโ data.py โ
โ โ โ โ โโโ evaluate.py โ
โ โ โ โ โโโ utils.py โ
โ โ โ โโโ configs/ โ
โ โ โ โ โโโ base.yaml # all params โ
โ โ โ โโโ notebooks/ โ
โ โ โ โโโ walkthrough.ipynb โ
โโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโ โโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโ
```
*[placeholder: animated GIF showing the full pipeline โ paper fetch โ parsing โ ambiguity audit โ code generation โ walkthrough notebook]*
---
## Why this exists
**The problem:** ML papers are vague. Critical hyperparameters are buried in appendices or omitted entirely. Prose contradicts equations. "Standard settings" refers to nothing specific. When you implement a paper, you spend more time detective-working than coding.
**What LLMs get wrong:** Naive code generation fills in every gap silently and confidently. You get something that runs but doesn't match the paper. Worse, you can't tell which parts are from the paper and which were invented by the model.
**What paper2code does differently:**
1. **Citation anchoring** โ every line of generated code references the exact paper section and equation it implements (`ยง3.2, Eq. 4`)
2. **Ambiguity auditing** โ before writing a single line of code, every implementation choice is classified as `SPECIFIED`, `PARTIALLY_SPECIFIED`, or `UNSPECIFIED`
3. **Honest uncertainty** โ unspecified choices are flagged with `[UNSPECIFIED]` comments at the exact line where the choice is made, with common alternatives listed
4. **Appendix mining** โ appendices, footnotes, and figure captions are treated as first-class sources, not ignored
The result: code you can trust because you can verify every decision against the paper.
---
## Install
```bash
npx skills add PrathamLearnsToCode/paper2code/skills/paper2code
```
You'll be prompted to:
1. **Select agents** โ pick the coding agents you want to use this skill with (e.g., Claude Code)
2. **Choose scope** โ Global (recommended) or project-level
3. **Choose method** โ Symlink (recommended) or copy
Once installed, open your agent and run the skill:
```bash
claude # or your preferred agent
```
---
## Usage
### Basic โ generate a minimal implementation
```
/paper2code https://arxiv.org/abs/1706.03762
```
### Specify framework
```
/paper2code https://arxiv.org/abs/2006.11239 --framework jax
```
### Full mode โ includes training loop and data pipeline
```
/paper2code 2106.09685 --mode full
```
### Educational mode โ extra comments and pedagogical notebook
```
/paper2code https://arxiv.org/abs/2010.11929 --mode educational
```
### Using bare arxiv ID
```
/paper2code 1706.03762
```
---
## What you get
```
attention_is_all_you_need/
โโโ README.md # Paper summary, contribution statement, quick-start
โโโ REPRODUCTION_NOTES.md # Ambiguity audit, unspecified choices, known deviations
โโโ requirements.txt # Pinned dependencies
โโโ src/
โ โโโ model.py # Architecture โ every layer cited to paper section
โ โโโ loss.py # Loss functions with equation references
โ โโโ data.py # Dataset class skeleton with preprocessing TODOs
โ โโโ train.py # Training loop (if in scope)
โ โโโ evaluate.py # Metric computation code
โ โโโ utils.py # Shared utilities (masking, positional encoding, etc.)
โโโ configs/
โ โโโ base.yaml # All hyperparams โ each one cited or flagged [UNSPECIFIED]
โโโ notebooks/
โโโ walkthrough.ipynb # Pedagogical notebook linking paper sections โ code โ sanity checks
```
### Key files explained
| File | Purpose |
|------|---------|
| `model.py` | Architecture only. Each class maps to a paper section. Variable names match paper notation. |
| `REPRODUCTION_NOTES.md` | The ambiguity audit. Lists every choice, whether the paper specified it, and what alternatives exist. |
| `base.yaml` | Single source of truth for all hyperparameters. |
| `walkthrough.ipynb` | Runnable on CPU with toy dimensions. Quotes paper passages, shows corresponding code, runs shape checks. |
---
## What this skill will NOT do
- **Won't guarantee correctness.** The implementation matches what the paper describes. If the paper is wrong, the code is wrong. If the paper is vague, the code flags it.
- **Won't invent details.** If the paper doesn't specify a hyperparameter, the code uses a common default and marks it `[UNSPECIFIED]`. It will never silently fill in gaps.
- **Won't download datasets.** The `data.py` provides a `Dataset` class skeleton with clear instructions on where to get the data and how to preprocess it.
- **Won't set up training infrastructure.** No distributed training, no experiment tracking, no checkpointing beyond what the paper's contribution requires.
- **Won't implement baselines.** Only the core contribution of the paper is implemented.
- **Won't reimplement standard components.** If the paper says "standard transformer encoder," the code imports it or notes the dependency โ it doesn't reimplement attention from scratch.
---
## Design principles
### Citation anchoring convention
Every non-trivial code decision is anchored to the paper:
```python
# ยง3.2 โ "We apply layer normalization before each sub-layer" (Pre-LN variant)
class TransformerBlock(nn.Module):
def forward(self, x):
# ยง3.2, Eq. 2 โ attention_weights = softmax(QK^T / sqrt(d_k))
attn_out = self.attention(self.norm1(x)) # (batch, seq_len, d_model)
x = x + attn_out # ยง3.2 โ residual connection
```
### The UNSPECIFIED flag system
```python
# [UNSPECIFIED] Paper does not state epsilon for LayerNorm โ using 1e-6 (common default)
# Alternatives: 1e-5 (PyTorch default), 1e-8 (some implementations)
self.norm = nn.LayerNorm(d_model, eps=1e-6)
```
```python
# [ASSUMPTION] Using pre-norm based on "we found pre-norm more stable" in ยง4.1
# The paper uses post-norm in Figure 1 but pre-norm in experiments โ ambiguous
```
### Ambiguity classification
| Tag | Meaning |
|-----|---------|
| `ยงX.Y` | Directly specified in paper section X.Y |
| `ยงX.Y, Eq. N` | Implements equation N from section X.Y |
| `[UNSPECIFIED]` | Paper does not state this โ our choice with alternatives listed |
| `[PARTIALLY_SPECIFIED]` | Paper mentions this but is ambiguous โ quote included |
| `[ASSUMPTION]` | Reasonable inference from paper context โ reasoning explained |
| `[FROM_OFFICIAL_CODE]` | Taken from the authors' official implementation |
---
## Contributing
### Adding worked examples
Worked examples are the most trust-building part of this project. To add one:
1. Pick a well-known paper (people should be able to verify the output)
2. Run the skill: `/paper2code https://arxiv.org/abs/XXXX.XXXXX`
3. Save the full output to `skills/paper2code/worked/{paper_slug}/`
4. Write a `review.md` that honestly evaluates:
- What the skill got right
- What it correctly flagged as unspecified
- Any mistakes it made
- Any edge cases it handled well or poorly
5. Submit a PR with all of the above
### Improving guardrails
If you find a pattern where the skill hallucinates or makes a silent assumption, add it to the appropriate file in `guardrails/`.
### Adding domain knowledge
If papers in your subfield consistently reference components that the skill doesn't know about (e.g., graph neural network primitives, RL components), add a knowledge file in `knowledge/`.
---
## Worked examples
This repo includes fully worked examples to demonstrate output quality:
| Paper | Type | Command |
|-------|------|---------|
| Attention Is All You Need (1706.03762) | Architecture | `/paper2code https://arxiv.org/abs/1706.03762` |
| DDPM (2006.11239) | Training method | `/paper2code https://arxiv.org/abs/2006.11239` |
Each includes the complete generated output plus an honest `review.md` evaluating what the skill got right and wrong.
---