314 skills · sorted by relevance
Analyze and modify refusal behavior in open-weight language models without retraining. Use OBLITERATUS to select methods suited to available hardware, compare approaches, and evaluate changes in refusal rates and response quality.
Verify RuView pipeline output against a reference hash, package evidence for others to check, and review accuracy statements. The skill distinguishes measured, claimed, and synthetic results and requires real device logs for hardware validation.
Set up RuView sensing nodes by building and flashing ESP32 firmware, configuring WiFi and the receiving address, and checking serial logs. Diagnose missing CSI streams, the WiFi signal measurements sent by sensing nodes.
Compress screen recordings on Apple Silicon Macs using dedicated video hardware. Choose quality settings for sharing or archiving, inspect video properties, and check the resulting file size and storage savings.
Compress screen recordings on Apple Silicon Macs using dedicated HEVC video encoding hardware. Choose quality presets for sharing or archiving, inspect source files, and review file size savings through command output.
Build and optimize quantum circuits with PennyLane, including models that combine quantum circuits with PyTorch or JAX. Explore molecular energy estimation and optimization problems, validate values and gradients in simulation, and prepare tested circuits for provider plugins.
Design custom laboratory parts as adjustable build123d models and export STEP, STL, and DXF fabrication files. Check mating dimensions against standards, test geometry, and visually review holders, mounts, microfluidic parts, and other research fixtures before fabrication.
Build and maintain an inventory of hardware, software, and cloud resources. Use guidance on automated discovery, ownership records, tagging checks, and integration with a configuration database to support operations and audit preparation.
Configure cryptographic key storage and operations through PKCS#11, a standard interface for security devices. Work with physical HSMs or SoftHSM2 for development to generate keys, sign data, encrypt information, and check key export restrictions.
Use Intel CHIPSEC to inspect firmware protections, including BIOS write locks and Secure Boot settings. Save test results, extract firmware for offline comparison, and review UEFI variables to investigate configuration weaknesses or suspicious changes.
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.
Read and control sensors, displays, and other peripherals on Sipeed boards through I2C and SPI, two hardware communication interfaces. Find connected devices, prepare shared pins, and troubleshoot wiring, permissions, and communication settings.
Turn an electronics or IoT idea into a working breadboard prototype. Get board and software recommendations, a plan for viewing sensor readings, and time and cost ranges tailored to experience, budget, and location.
Investigate slow TPU training on GKE by comparing diagnostic events with one-minute monitoring metrics. Separate memory or CPU bottlenecks from hardware and network issues, and identify the appropriate next steps.
Investigate stalled TPU training across multiple slices on GKE using hang reports, logs, and latency metrics. Separate compilation inconsistencies and data input stalls from hardware or network faults, and identify the relevant nodes for further investigation.
Diagnose unresponsive GKE GPU or TPU nodes and pods stuck terminating through read-only checks of node status, logs, serial output, and repair history. Receive documented recovery guidance to apply yourself.
Deploy Model Garden models to Agent Platform endpoints, the addresses used for model requests. Check supported hardware and estimated hourly costs, inspect deployment progress, test predictions, and remove deployed resources after use.
Plan and troubleshoot large language and vision model training and inference. This guide uses hardware measurements to identify bottlenecks, address memory shortages and unstable training, and improve checkpointing, recovery, and performance across multiple GPUs and machines.
Rent and manage vast.ai GPUs based on the training task rather than a chosen hardware model. The skill estimates requirements, compares offers by total expected cost, prepares the selected instance, and retrieves results before destroying it.
Add new operators to MNN through a staged workflow covering data definitions, output shapes, computation, and tests. It prioritizes combining existing operators, with CPU implementations and additional hardware backends used when needed.
Guides LLM compression through pruning, which removes selected model weights. Covers Wanda and SparseGPT methods without retraining, accuracy comparisons, and structured weight patterns that can accelerate inference on compatible hardware.
Implement Mamba, a model architecture that processes sequences through an internal state with computation scaling linearly with input length. Covers pretrained text generation, Mamba-1 and Mamba-2 configuration, speed comparisons with Transformers, and GPU memory troubleshooting.
Fine-tune vision models on custom images using Hugging Face Jobs cloud GPUs. Covers object detection, image classification, and segmentation—identifying image regions—with dataset validation, hardware selection, cost estimates, monitoring, and saving trained models to the Hub.
Build and maintain machine learning apps on Hugging Face Spaces. Choose a suitable app framework and computing hardware, load models, deploy files, and check logs and outputs to diagnose build or runtime problems.