haddock-development/ttc-pi-package/ttc-distill

Claude Code and agent skill by haddock-development/ttc-pi-package

Trust tier 3
  • community
Source check pending

Checking that the SKILL.md file is still in the repository.

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.

Install

npx vskill@latest install haddock-development/ttc-pi-package/ttc-distill

SKILL.md

TTC Distillation Skill

Distill knowledge from large teacher models to smaller student models while preserving TTC capabilities.

Setup

bash
# 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')"

Distillation Profiles

Medium Profile (Free)
  • Backend: Kaggle GPU (T4/P100)
  • Batch Size: 4-8
  • Teacher: 7B-14B models
  • Time: 2-4 hours
  • Cost: Free (30h/week limit)
bash
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
High Profile (Paid)
  • Backend: HF Jobs (A10G/L4)
  • Batch Size: 16-32
  • Teacher: 14B-72B models
  • Time: 4-12 hours
  • Cost: ~$5-20
bash
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

Distillation Pipeline

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

Input Contract

  • teacher_model_id: HuggingFace model ID
  • student_model_id: Base student model ID
  • dataset: Path to distillation dataset
  • profile: "medium" or "high"

Output Contract

  • distill_summary.json: Training metrics
  • final_model/: Distilled student model
  • teacher_traces.jsonl: Teacher reasoning traces (optional)

Example

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

Quality Metrics

After distillation, check:

  • Perplexity on validation set (< 5.0 is good)
  • Teacher alignment (> 80% agreement)
  • Benchmark scores (MMLU, GSM8K, etc.)

Security

  • Security scanPassed, 90 of 100Pattern checks on the skill's files.
  • Model reviewPassed

See the full security report

Quality evaluation

  • Task evaluationNot runNo evaluation yet. Run one in Skill Studio.

See evaluation results

Source

Popularity

7-day trend score
0

Not refreshed yet.

Works with

Any agent vskill installs to: 56 targets in all. These 39 have a page on this site.

  • Claude Code
  • Cursor
  • GitHub Copilot
  • Windsurf
  • Codex
  • Gemini CLI
  • Cline
  • Amp
  • Roo Code
  • Goose
  • Continue
  • Kilo Code
  • Junie
  • Trae
  • OpenHands
  • Qwen Code
  • AI Agent
  • Tabnine Chat
  • Sourcegraph Cody
  • Amazon Q
  • Firebase Genkit
  • Sweep
  • Devin
  • Phind
  • Mentat
  • Claude Desktop
  • Aider
  • GPT Pilot
  • Plandex
  • bolt.new
  • v0
  • Replit Agent
  • Anse
  • TypingMind
  • BetterChatGPT
  • LibreChat
  • Warp AI
  • Pieces
  • Fig AI
Embed badge
vskill VERIFIED badge
[![vskill VERIFIED](https://verified-skill.com/api/v1/skills/haddock-development/ttc-pi-package/ttc-distill/badge)](https://verified-skill.com/skills/haddock-development/ttc-pi-package/ttc-distill)

Something wrong with this skill or its page? Tell us.