Toolverse
All skills

knowledge-distillation

by davila7

Compress large language models using knowledge distillation from teacher to student models. Use when deploying smaller models with retained performance, transferring GPT-4 capabilities to open-source models, or reducing inference costs. Covers temperature scaling, soft targets,

Installation

Pick a client and clone the repository into its skills directory.

Installation

Quick info

Author
davila7
Category
Security

About this skill

Compress large language models using knowledge distillation from teacher to student models. Use when deploying smaller models with retained performance, transferring GPT-4 capabilities to open-source models, or reducing inference costs. Covers temperature scaling, soft targets, reverse KLD, logit distillation, and MiniLLM training strategies.

How to use

  1. Zainstaluj wymagane biblioteki: pip install transformers datasets accelerate torch deepspeed wandb. Opcjonalnie sklonuj repozytorium MiniLLM z GitHub (microsoft/LMOps), aby uzyskać zaawansowane implementacje destylacji.

  2. Przygotuj parę modeli: załaduj duży model nauczycielski (np. Llama-2-70b) i mały model ucznia (np. Llama-2-7b) za pomocą AutoModelForCausalLM z biblioteki transformers.

  3. Wybierz technikę destylacji odpowiednią do Twojego przypadku użycia: skalowanie temperatury dla prostszych scenariuszy, destylację logitów dla transferu wiedzy z modeli proprietarnych, lub odwrotną dywergencję KL (MiniLLM) dla zaawansowanej kompresji.

  4. Przygotuj dane treningowe — możesz użyć danych syntetycznych wygenerowanych przez model nauczycielski lub istniejące zbiory danych dostosowane do Twojej domeny.

  5. Skonfiguruj parametry treningu w TrainingArguments (liczba epok, rozmiar batcha, współczynnik uczenia) i uruchom trening za pomocą klasy Trainer, monitorując postęp w Weights & Biases.

  6. Po treningu przetestuj model ucznia na zadaniach docelowych, aby potwierdzić, że zachował co najmniej 90% wydajności oryginalnego modelu nauczycielskiego przy znacznie mniejszym rozmiarze.

Related skills

reviewing-code

by CaptainCrouton89

Systematically evaluate code changes for security, correctness, performance, and spec alignment. Use when reviewing PRs, assessing code quality, or verifying implementation against requirements.

Security
1493

llama-cpp

by zechenzhangAGI

Runs LLM inference on CPU, Apple Silicon, and consumer GPUs without NVIDIA hardware. Use for edge deployment, M1/M2/M3 Macs, AMD/Intel GPUs, or when CUDA is unavailable. Supports GGUF quantization (1.5-8 bit) for reduced memory and 4-10× speedup vs PyTorch on CPU.

Security
11252

zendesk

by vm0-ai

Zendesk Support REST API for managing tickets, users, organizations, and support operations. Use this skill to create tickets, manage users, search, and automate customer support workflows.

Security
11100

security-compliance

by davila7

Guides security professionals in implementing defense-in-depth security architectures, achieving compliance with industry frameworks (SOC2, ISO27001, GDPR, HIPAA), conducting threat modeling and risk assessments, managing security operations and incident response, and embedding

Security
1172

accessibility-compliance

by wshobson

Implement WCAG 2.2 compliant interfaces with mobile accessibility, inclusive design patterns, and assistive technology support. Use when auditing accessibility, implementing ARIA patterns, building for screen readers, or ensuring inclusive user experiences.

Security
2173

python-expert

by Shubhamsaboo

Senior Python developer expertise for writing clean, efficient, and well-documented code.\nUse when: writing Python code, optimizing Python scripts, reviewing Python code for best practices,\ndebugging Python issues, implementing type hints, or when user mentions Python, PEP 8,

Security
2777