菜单
Reproduce experiments from academic papers — parse paper, setup environment, find datasets, generate code, run and verify.
用 Codex 或 Claude 帮你安装 复制这段 Prompt，粘贴到 Codex、Claude 或其他助手里，让它检查 Skill 页面并帮你完成安装。
基于 SOC 职业分类
Search HuggingFace Hub for datasets by keyword. Use when the user wants to find training data, benchmarks, or evaluation datasets for ML/NLP/CV research.
Fetch the FULL TEXT of an arXiv paper (all sections — introduction, method, results, conclusion). Use when you need to read beyond the abstract into the paper's actual content. Only works for arXiv papers. For metadata/abstract only, use paper-read. For local PDFs, use paper-read-pdf.
Two-layer memory system with grep-based recall for research sessions.
Track specific scientific claims across the literature over time — who made it, who replicated it, who challenged it, whether it still stands. Use when verifying a key assumption before building on it, or when checking whether a published result has been updated or superseded.
Scan papers for conflicting empirical claims, methodological disagreements, or opposing conclusions on the same topic. Use when writing discussion sections, evaluating conflicting results, or checking if a claim is contested before building on it.
Synthesize findings across multiple papers into a coherent narrative, structured comparison table, or temporal evolution. Use after collecting papers via survey or paper-search. Goes beyond summarizing individual papers to produce insights that only emerge when reading across the corpus as a whole.
| name | reproduce |
| description | Reproduce experiments from academic papers — parse paper, setup environment, find datasets, generate code, run and verify. |
| always | false |
| requires | {"bins":[],"env":[]} |
paper_fetch — Download full paper text from arXiv (HTML/abstract)paper_read — Read paper metadata from Semantic Scholarweb_search — Search for official code repos, datasetsweb_fetch — Fetch GitHub repos, Papers With Code pagesdataset_search — Search HuggingFace Hub for datasetsdataset_download — Generate dataset download scriptenv_builder — Generate Dockerfile, requirements.txt, setup scriptscode_gen — Scaffold full training project (model/train/eval)exec — Run scripts in current environmentwrite_file — Write any fileread_file — Read files1. paper_fetch(paper_id="1706.03762") → full text
2. paper_read(paper_id="1706.03762") → metadata, citations
3. LLM extracts structured spec:
{
architecture: "Transformer, 6 encoder+decoder layers, d_model=512, h=8",
hyperparams: {lr: 0.0001, batch_size: 32, epochs: 100, optimizer: "adam", warmup: 4000},
dataset: "WMT 2014 English-German",
metrics: [{name: "BLEU", value: 28.4}],
hardware: "8x P100 GPUs, 3.5 days"
}
4. web_search("{paper_title} github code") → find official repo
5. web_fetch("https://paperswithcode.com/api/v1/papers/?q={title}")
→ structured code links
6. dataset_search(query="{dataset_name}") → HuggingFace match
7. Create project dir: outputs/reproduction_{arxiv_id}/
8. env_builder(
output_dir="outputs/reproduction_{arxiv_id}",
framework="pytorch",
packages=["transformers", "datasets", "sacrebleu"]
) → Dockerfile, requirements.txt, setup.bat/.sh
9. exec("pip install -r outputs/reproduction_{arxiv_id}/requirements.txt")
10. dataset_download(
dataset_name="{dataset_name}",
output_dir="outputs/reproduction_{arxiv_id}",
source="auto"
) → download_data.py
11. exec("python outputs/reproduction_{arxiv_id}/download_data.py")
If official repo found:
12. exec("git clone {repo_url} outputs/reproduction_{arxiv_id}/src_original")
13. Read src files, identify entry point
14. Fix deprecated APIs, import errors with write_file + exec
If no repo:
12. code_gen(
output_dir="outputs/reproduction_{arxiv_id}",
paper_title="{title}",
framework="pytorch",
architecture="{architecture}",
hyperparams={...},
dataset_name="{dataset_name}",
task="seq2seq"
) → model.py, train.py, dataset.py, evaluate.py
13. Implement TODOs in model.py and dataset.py based on paper text
(LLM fills in architecture details using paper_fetch content)
14. exec("python outputs/reproduction_{arxiv_id}/src/train.py --config configs/default.yaml")
→ monitor for errors
15. On error: read traceback → fix → retry (max 3 times)
16. exec("python outputs/reproduction_{arxiv_id}/src/evaluate.py --checkpoint ...")
17. Compare metrics vs paper-reported values
18. write_file("outputs/reproduction_{arxiv_id}/REPRODUCTION_REPORT.md", content="""
## Paper: {title}
## Status: SUCCESS / PARTIAL / FAILED
## Metric comparison table
## Notes on differences
""")
outputs/reproduction_{arxiv_id}/
├── configs/
│ └── default.yaml ← hyperparameters from paper
├── src/
│ ├── model.py ← architecture
│ ├── train.py ← training loop
│ ├── dataset.py ← data loading
│ └── evaluate.py ← evaluation
├── data/ ← downloaded dataset
├── outputs/
│ ├── checkpoints/ ← saved model weights
│ └── logs/ ← training logs
├── Dockerfile
├── requirements.txt
├── setup.sh / setup.bat
├── download_data.py
├── README.md
└── REPRODUCTION_REPORT.md ← comparison vs paper
| Level | Criteria |
|---|---|
| ✅ Level 1 | Code runs without errors |
| ✅ Level 2 | Loss decreases, training converges |
| ✅ Level 3 | Metrics within ±5% of paper |
| ✅ Level 4 | Metrics within 1 std dev (multi-seed) |
| 🏆 Level 5 | Bit-for-bit exact match (rare) |
Always inject into generated code:
import random, numpy as np, torch, os
os.environ["PYTHONHASHSEED"] = "0"
os.environ["CUBLAS_WORKSPACE_CONFIG"] = ":4096:8"
random.seed(seed)
np.random.seed(seed)
torch.manual_seed(seed)
torch.cuda.manual_seed_all(seed)
torch.backends.cudnn.deterministic = True
torch.use_deterministic_algorithms(True, warn_only=True)