skillZs
★ LIVE SKILL TAGS ★
>>> LIVE SKILLS INDEX <<<
* OPEN SOURCE *
NO LOGIN, NO TRACKING
※ REAL INSTALL DATA ※
← back to all skills
eyadsibai/ltk79 installs

llm-training

Use when "training LLM", "finetuning", "RLHF", "distributed training", "DeepSpeed", "Accelerate", "PyTorch Lightning", "Ray Train", "TRL", "Unsloth", "LoRA training", "flash attention", "gradient checkpointing"

How do I install this agent skill?

npx skills add https://github.com/eyadsibai/ltk --skill llm-training
view source ↗

Is this agent skill safe to install?

  • Gen Agent Trust Hubpass

    The skill provides informational content and framework comparisons for LLM training and finetuning. It does not contain any executable code, malicious instructions, or security risks.

  • Socketpass

    No alerts

  • Snykpass

    Risk: LOW · No issues

  • Runlayerwarn

    1/1 file flagged

What does this agent skill do?

LLM Training

Frameworks and techniques for training and finetuning large language models.

Framework Comparison

FrameworkBest ForMulti-GPUMemory Efficient
AccelerateSimple distributedYesBasic
DeepSpeedLarge models, ZeROYesExcellent
PyTorch LightningClean training loopsYesGood
Ray TrainScalable, multi-nodeYesGood
TRLRLHF, reward modelingYesGood
UnslothFast LoRA finetuningLimitedExcellent

Accelerate (HuggingFace)

Minimal wrapper for distributed training. Run accelerate config for interactive setup.

Key concept: Wrap model, optimizer, dataloader with accelerator.prepare(), use accelerator.backward() for loss.


DeepSpeed (Large Models)

Microsoft's optimization library for training massive models.

ZeRO Stages:

  • Stage 1: Optimizer states partitioned across GPUs
  • Stage 2: + Gradients partitioned
  • Stage 3: + Parameters partitioned (for largest models, 100B+)

Key concept: Configure via JSON, higher stages = more memory savings but more communication overhead.


TRL (RLHF/DPO)

HuggingFace library for reinforcement learning from human feedback.

Training types:

  • SFT (Supervised Finetuning): Standard instruction tuning
  • DPO (Direct Preference Optimization): Simpler than RLHF, uses preference pairs
  • PPO: Classic RLHF with reward model

Key concept: DPO is often preferred over PPO - simpler, no reward model needed, just chosen/rejected response pairs.


Unsloth (Fast LoRA)

Optimized LoRA finetuning - 2x faster, 60% less memory.

Key concept: Drop-in replacement for standard LoRA with automatic optimizations. Best for 7B-13B models.


Memory Optimization Techniques

TechniqueMemory SavingsTrade-off
Gradient checkpointing~30-50%Slower training
Mixed precision (fp16/bf16)~50%Minor precision loss
4-bit quantization (QLoRA)~75%Some quality loss
Flash Attention~20-40%Requires compatible GPU
Gradient accumulationEffective batch↑No memory cost

Decision Guide

ScenarioRecommendation
Simple finetuningAccelerate + PEFT
7B-13B modelsUnsloth (fastest)
70B+ modelsDeepSpeed ZeRO-3
RLHF/DPO alignmentTRL
Multi-node clusterRay Train
Clean code structurePyTorch Lightning

Resources

Add the canonical catalog link to the repository README so users can inspect current installs and available audits. The publishing guide covers the complete discovery path.

<a href="https://skillzs.dev/skills/eyadsibai/ltk/llm-training">View llm-training on skillZs</a>