skills.rest
VKirill
VKirillCOMMUNITY
Explore all skills by this author →
GitHub

transformers

Load and run HuggingFace Transformers models for inference and fine-tuning.

15|2|Updated May 23, 2026
Search Tags:#fine-tuning#quantization#lora#huggingface#transformers#text generation#model inference
One-click install
npx skills add https://github.com/VKirill/antigravity-for-claude-code --skill transformers-vkirill
Or copy as Structured Prompt for Agent▼
Please help me install this Agent Skill.
Skill: transformers
Source: https://github.com/VKirill/antigravity-for-claude-code/tree/main/skills/transformers
Command: npx skills add https://github.com/VKirill/antigravity-for-claude-code --skill transformers-vkirill

SYSTEM DOCUMENTATION & REQUIREMENTS

💡 This Skill includes references (resource) components.

What problem does it solve?

Transformers helps you reliably load HuggingFace pretrained models and run inference or fine-tuning without getting stuck on tokenizer/template mismatches, generation settings, or GPU/quantization configuration.

Core Features & Use Cases

  • Model loading & task routing: Use AutoTokenizer and AutoModelFor* to load the correct architecture from a single checkpoint string.
  • Inference & generation: Use pipeline() or model.generate() with correct max_new_tokens, EOS/PAD handling, and optional streaming via TextIteratorStreamer.
  • Training & adaptation: Fine-tune with Trainer/TrainingArguments, including PEFT LoRA/QLoRA and bitsandbytes quantization for memory-efficient runs.
  • Data & multimodal support: Tokenize datasets with datasets.load_dataset + .map(), and handle vision/audio with AutoProcessor and appropriate multimodal pipelines.

Quick Start

Use this skill to fine-tune and generate with a consistent HF stack by running the Transformers flow for the task and checkpoint you need.

Frequently Asked Questions about transformers

High-intent search queries and answers about installing and using this skill.

FAQPage Schema
How do I load and run a HuggingFace Transformers model for inference?▼

To load and run HuggingFace Transformers models, use AutoTokenizer and AutoModelFor* with pipeline() for quick inference, or model.generate() for controlled decoding. Ensure correct checkpoint-aligned tokenizer and model pairing to avoid template mismatches.

What is the best way to fine-tune a HuggingFace model with LoRA and quantization?▼

Fine-tuning HuggingFace models with LoRA and quantization uses Trainer and TrainingArguments with PEFT LoRA/QLoRA alongside bitsandbytes. This enables memory-efficient adaptation while configuring device_map, dtype, and attention settings correctly.

How do I tokenize datasets safely for HuggingFace model training?▼

Tokenize datasets safely for HuggingFace training by loading data with datasets.load_dataset and applying preprocessing via datasets.map using the processing_class. Follow batching padding_side rules to ensure correct input formatting for the model.

Does HuggingFace Transformers support vision and audio multimodal tasks?▼

HuggingFace Transformers supports vision and audio multimodal tasks by using AutoProcessor and appropriate multimodal pipelines. It routes text, vision, and audio inputs through the correct architecture loaded from a single checkpoint string.

Why does my HuggingFace text generation output repeat or stop prematurely?▼

HuggingFace text generation issues like repetition or premature stopping often stem from incorrect EOS and PAD token handling in model.generate(). Configure max_new_tokens carefully and ensure tokenizer generation settings align with the model checkpoint.

Related Skills

Explore More Software Engineering Skills→
shushuznshushuzn
4

transformers

Run transformer model inference and fine-tuning across NLP, vision, and audio tasks.

Community
Intermediate
ScientiaCapitalScientiaCapital
1

model-deployment

Export fine-tuned models to GGUF and deploy with vLLM or Ollama.

Community
Advanced
rnbenrnben

peft-fine-tuning

Fine-tune 7B-70B models with LoRA and QLoRA adapters using PEFT.

Community
Advanced
Clay-HHKClay-HHK
2

optimizing-attention-flash

Optimize transformer attention to reduce GPU memory and accelerate computation.

Community
Advanced