734 skills · sorted by relevance
Plan memory use and monitor long training jobs on NVIDIA DGX Spark. It guides troubleshooting for out-of-memory failures, tracks temperature and power alongside training logs, and helps assess competition between training and inference workloads.
Add Arize AX tracing to an LLM application, recording how model calls and tool executions connect. Inspect the existing code, select suitable integrations, add instrumentation, and verify that the application runs and traces reach Arize.
Investigate SGLang runs that freeze across multiple GPUs. It uses stack traces, communication logs, GPU memory dumps, and comparisons between processes to locate state differences, trace their causes, and guide fixes.
Research private companies through six perspectives, including business, finances, competition, and technology. Cross-check evidence, separate facts from inferences, and produce a valuation range, possible exit routes, and a map of missing information.
Reduce the memory needed to run large language models with GPTQ quantization, which stores model weights at lower precision. Covers loading compressed models, quantizing your own models, choosing execution backends, and connecting to memory-efficient fine-tuning.
Run machine learning workloads on Modal's cloud GPUs without managing servers. Guides Python configuration for training, inference, batch processing, and API deployment, including automatic scaling, persistent storage, scheduled jobs, and troubleshooting.
Build and evaluate Mamba models for long sequences using state-space models, which summarize past inputs in an internal state. Covers setup, pretrained text generation, differences between Mamba-1 and Mamba-2, and speed comparisons with Transformers.
Reduce GPU memory use by loading language models in 8-bit or 4-bit formats with bitsandbytes and HuggingFace Transformers. Includes configuration examples, QLoRA fine-tuning of compressed models, and optimizers that reduce memory needed during training.
Guides the use of RWKV, a language model architecture that carries information forward in a compact state. Covers streaming text generation, long document processing, fine-tuning, and memory comparisons with Transformers.
Deploy language models with vLLM through an OpenAI-compatible API or offline batch processing. Tune response speed, throughput, and GPU memory use with quantization, which reduces numerical precision, and execution across multiple GPUs.
Build data processing pipelines for machine learning, from preprocessing to batch predictions. Uses Ray Data to handle datasets larger than memory, distribute work across CPUs and GPUs, and feed data into training frameworks.
Build table data processing with TanStack Table v9 while keeping data and column definitions stable. The skill guides type inference, which determines types automatically, row identity, optional features, and checks for missing APIs.
Build visual AI systems that detect objects or divide images into labeled regions. This skill guides dataset preparation, model selection, training, evaluation, inference optimization, and deployment using frameworks such as PyTorch and Ultralytics.
Find official NemoClaw documentation for installation, configuration, operation, and troubleshooting. Get source-linked guidance tailored to the selected operating system, agent variant, and inference provider, with optional documentation MCP setup when requested.
Validate the staging NemoClaw Brev Launchable by checking its web deployment flow, environment access, deployed image, and preinstalled user journey. Report inference results and missing dependencies separately as advisory evidence outside automated end-to-end testing.
Plan and create Google Kubernetes Engine clusters using templates for production, development, or AI workloads. Compare Google-managed and self-managed node options, review configurations before creation, and check settings for production readiness.
Manage isolated working environments in OpenShell using its command-line tool. Register gateways, create and connect to sandboxes, configure credential providers, and reuse workload templates with shared images, environment settings, and computing resources.
Diagnose errors and incorrect inference results in MNN's QNN backend for Qualcomm HTP processors. Compare intermediate calculation data against CPU results to locate the first faulty operator, distinguish bugs from precision limits, and guide operator support and performance fixes.
Improve MNN inference on OpenCL by measuring bottlenecks and refining GPU kernels, the code that performs calculations. The skill guides baseline measurements, correctness checks, Android device benchmarks, and integration of new OpenCL features with compatibility fallbacks.
Deploy language model APIs and run batch generation with vLLM. Configure GPU memory use, quantization—lower precision model storage—and distribution across GPUs, then test response times and monitor how many requests the server handles.
Guides adapting OpenPI robot models to custom datasets and serving them to produce actions from observations. Covers JAX and PyTorch training, checkpoint conversion, and troubleshooting for normalization statistics and GPU memory.
Guides fine-tuning and evaluation of OpenVLA-OFT robot models that turn images and language into actions. Covers LIBERO simulation, ALOHA deployment, LoRA adaptation that updates a small set of parameters, and troubleshooting action scaling or configuration mismatches.
Build data processing pipelines for machine learning, from preprocessing to batch predictions. Uses Ray Data to handle datasets larger than memory, distribute work across CPUs and GPUs, and feed data into training frameworks.
Optimize transformer attention, the calculation that relates tokens to one another, using Flash Attention. The skill guides PyTorch integration, GPU memory reduction, performance measurement, and output comparisons for training or inference with long sequences.