Skill UI

Orchestrates PyTorch-native TorchTitan 4D parallelism (FSDP2, TP, PP, CP) for large-scale LLM pretraining, scaling from 8 to 512+ GPUs with Float8, torch.compile, and distributed checkpointing to boot Llama 3.1, DeepSeek V3, or custom models.

GPTQ enables post-training 4-bit quantization for large LLMs, delivering up to 4× memory reduction and 3-4× faster inference with under 2% perplexity loss on consumer GPUs through AutoGPTQ, PEFT, and Transformer integrations.

Execute complex and compute-intensive workloads directly on fully managed Hugging Face cloud infrastructure. This skill eliminates the need for local setup, allowing users to run jobs using cloud CPUs, GPUs, or TPUs. Ideal for large-scale data processing, batch inference, reproducible model training, and scheduled automation tasks, while persisting results to the Hugging Face Hub.

Playbook for running ML training and inference on Lambda Labs’ on-demand GPU cloud, covering when to choose it, console/SSH/API workflows, available GPUs, pricing, and persistent storage with Lambda Stack pre-installed.

llama.cpp is a pure C/C++ library designed for efficient LLM inference, optimized to run on resource-constrained environments. It excels in scenarios where high-end NVIDIA GPUs (CUDA) are unavailable, making it ideal for Apple Silicon Macs, AMD/Intel GPUs, or edge deployment devices like Raspberry Pi. It supports GGUF quantization formats, enabling significant memory reduction and speed improvements for local, cross-platform AI applications.

Modal Serverless GPU provides on‑demand, pay‑per‑second access to T4 through H200 GPUs, letting you deploy ML models as auto‑scaling APIs, run batch jobs, and prototype without managing infrastructure.

NVIDIA’s NeMo Guardrails adds runtime safety rails to LLM apps, covering jailbreak detection, hallucination and fact-checking, PII filtering, and toxicity checks via Colang 2.0 rules and integrations for production use on T4 GPUs.

Guides researchers through nnsight’s tracing and activation APIs so you can inspect or modify PyTorch internals locally or via NDIF-powered remote runs on 70B+ foundation models without local GPUs.

Covers HuggingFace PEFT workflows for LoRA/QLoRA-based LLM fine-tuning, enabling sub-1% parameter updates on limited GPUs plus adapter serving tips.

Optimizes large language model inference on NVIDIA GPUs with TensorRT, delivering 10-100× faster throughput, quantized precision (FP8/INT4), multi-GPU scaling, and serving-ready tooling for production deployments.

torchforge is Meta's advanced PyTorch-native library designed for agentic Reinforcement Learning (RL). It fundamentally separates core RL algorithms from complex distributed infrastructure concerns. This allows researchers to rapidly prototype and experiment with novel RL methods (like GRPO) using clean, PyTorch-native abstractions, while automatically handling large-scale tasks like distributed training across thousands of GPUs using Monarch and TorchTitan.

This guide demonstrates how to implement robust Role-Based Access Control (RBAC) and financial governance for cloud GPU environments. It allows organizations to allocate resources, set team-specific budgets, whitelist approved GPUs, and track spending via detailed audit logs. Ideal for multi-team research and engineering environments requiring strict cost accountability and access separation.