القائمة
TTC distillation skill for knowledge transfer from teacher to student models. Use when distilling large models to smaller ones with TTC awareness. Supports medium and high profiles.
التثبيت باستخدام Codex أو Claude انسخ هذا Prompt والصقه في Codex أو Claude أو مساعد آخر ليراجع صفحة Skill ويثبّتها لك.
استنادا إلى تصنيف SOC المهني
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.
Hyperparameter tuning skill for TTC training using Optuna. Use when optimizing learning rate, batch size, LoRA rank, and other training parameters.
Judge model training skill for TTC evaluation. Use when training models to evaluate and score outputs from other models. Supports reward model training and judge datasets.
TTC router skill for inference scaling decisions. Use when training, evaluating, or benchmarking the router that decides between low/medium/high/superthink compute modes.
TTC-aware training skill for Qwen models. Use when training models with Test-Time Compute awareness, including CPT → SFT → RLVR → TTC pipeline stages. Supports Kaggle and HF Jobs backends.
| name | ttc-distill |
| description | TTC distillation skill for knowledge transfer from teacher to student models. Use when distilling large models to smaller ones with TTC awareness. Supports medium and high profiles. |
| compatibility | Requires access to teacher model and compute backend |
Distill knowledge from large teacher models to smaller student models while preserving TTC capabilities.
# Ensure HF token has access to teacher model
huggingface-cli login
# Verify teacher model access
python -c "from transformers import AutoModel; AutoModel.from_pretrained('teacher-model-id')"
python scripts/distill.py \
--teacher Qwen/Qwen2.5-7B-Instruct \
--student Qwen/Qwen2.5-0.5B \
--dataset path/to/distill_data \
--profile medium \
--backend kaggle
python scripts/distill.py \
--teacher Qwen/Qwen2.5-72B-Instruct \
--student Qwen/Qwen3.5-0.8B \
--dataset path/to/distill_data \
--profile high \
--backend hf-jobs
1. Load teacher model (frozen)
2. Load student model (trainable)
3. Generate teacher outputs on dataset
4. Train student to match teacher
5. Evaluate distillation quality
6. Save student model
teacher_model_id: HuggingFace model IDstudent_model_id: Base student model IDdataset: Path to distillation datasetprofile: "medium" or "high"distill_summary.json: Training metricsfinal_model/: Distilled student modelteacher_traces.jsonl: Teacher reasoning traces (optional)User: Distill Qwen 2.5 7B to 0.8B using my math dataset
Agent will:
1. Check teacher access
2. Prepare distillation config
3. Launch job on selected backend
4. Monitor progress
5. Return distilled model path
After distillation, check: