| name | paper-to-engineering-features |
| description | Transform HuggingFace/arXiv research papers into trainable engineering features, skills, scripts and configs for Qwen 3.5 0.8B. Use when user provides a paper URL or wants to convert research into implementation. Creates complete skill ecosystems with ASCII diagrams. |
Paper to Engineering Features
Converts research papers into trainable engineering features for the TTC Stack.
Input
User provides a paper URL:
- HuggingFace Papers:
https://huggingface.co/papers/<paper-id>
- arXiv:
https://arxiv.org/abs/<paper-id>
Workflow
1. Fetch Paper
HuggingFace Papers:
PAPER_ID="<paper-id from URL>"
curl -s "https://huggingface.co/api/papers/$PAPER_ID"
arXiv:
PAPER_ID="<paper-id from URL>"
curl -s "http://export.arxiv.org/api/query?id_list=$PAPER_ID"
2. Extract Engineering Features
| Category | What to Look For |
|---|
| Architecture | Layer modifications, attention variants, skip connections, MoE |
| Training | Loss functions, optimization, curriculum learning |
| Algorithms | Sampling strategies, routing logic, decoding |
| Data | Dataset requirements, preprocessing |
| Evaluation | Benchmarks, metrics, baselines |
3. Map to TTC Stages
CPT (Continued Pre-Training)
├── Architecture changes → checkpoint surgery
└── Tokenizer changes → vocabulary expansion
SFT (Supervised Fine-Tuning)
├── Instruction format → chat template configs
└── Data requirements → dataset schemas
RLVR (Reinforcement Learning)
├── Reward signals → verifier configs
└── Policy updates → GRPO/DPO configs
TTC (Test-Time Compute)
├── Router decisions → router configs
└── Parallel sampling → best-of-N configs
4. Generate Skill Package
qwen-<feature-name>/
├── SKILL.md
├── scripts/
│ ├── train.py
│ ├── evaluate.py
│ └── inference.py
├── references/
│ ├── architecture.md
│ └── paper-summary.md
└── configs/
├── sft.yaml
├── rlvr.yaml
└── router.yaml
5. Create ASCII Diagram
┌─────────────────────────────────────────────┐
│ Qwen 3.5 0.8B │
│ ┌─────────────────────────────────────┐ │
│ │ [Feature modifications] │ │
│ └─────────────────────────────────────┘ │
└─────────────────────────────────────────────┘
6. Create GitHub Repos
gh repo create qwen-<feature-name> --public --description "Qwen implementation"
git init && git add . && git commit -m "Initial skill ecosystem"
git push origin main
Output
Provide:
- Paper Summary (3 bullets)
- Features Table
- TTC Stage Mapping
- Generated Skills (3-5 per paper)
- ASCII Diagram
- GitHub Repo Links
- Installation Commands
Example
From https://huggingface.co/papers/2512.24880:
| Skill | Purpose | Backend |
|---|
| qwen-mhc-architecture-lab | Architecture retrofit | Kaggle |
| qwen-mhc-kaggle-smoke | Quick validation | Kaggle |
| qwen-mhc-hf-jobs | Full training | HF Jobs |
| qwen-mhc-bridge | Pipeline bridge | N/A |
| qwen-mhc-retrofit-lab | Checkpoint surgery | Local |