“mechanistic interpretability”的搜索结果

Investigate how transformer language models produce predictions with TransformerLens. This skill guides inspection of intermediate values and attention patterns, plus experiments that replace internal values to test their effects and identify components involved in model behavior.

中文简介正在准备中。当前显示英语简介。

• Inspecting attention patterns and intermediate values
• Testing output changes by replacing internal activations
• Analyzing attention head contributions to predictions

Investigate how transformer language models produce predictions with TransformerLens. This skill guides inspection of intermediate values and attention patterns, plus experiments that replace internal values to test their effects and identify components involved in model behavior.

中文简介正在准备中。当前显示英语简介。

• Inspecting attention patterns and intermediate values
• Testing output changes by replacing internal activations
• Analyzing attention head contributions to predictions

Investigate how transformer language models produce predictions with TransformerLens. This skill guides inspection of intermediate values and attention patterns, plus experiments that replace internal values to test their effects and identify components involved in model behavior.

中文简介正在准备中。当前显示英语简介。

• Inspecting attention patterns and intermediate values
• Testing output changes by replacing internal activations
• Analyzing attention head contributions to predictions

Guides experiments that inspect and change internal model values using nnsight. Covers local PyTorch models and remote experiments through NDIF, including replacing intermediate values to test how particular layers affect a model’s predictions.

中文简介正在准备中。当前显示英语简介。

• Inspecting layer outputs and attention patterns
• Replacing internal values to compare predictions
• Running large-model experiments through NDIF

Guides experiments that inspect and modify internal values in PyTorch models to study their effects on predictions. Covers local analysis and remote experiments through NDIF for models too large for local GPUs.

中文简介正在准备中。当前显示英语简介。

• Inspect internal values and attention patterns across layers
• Replace internal values to compare changes in predictions
• Run large-model experiments without local GPUs

Provides guidance for mechanistic interpretability research using TransformerLens to inspect and manipulate transformer internals via HookPoints and activation caching. Use when reverse-engineering model algorithms, studying attention patterns, or performing activation patching experiments.

中文简介正在准备中。当前显示原文说明。

Fornece orientação para interpretar e manipular internals de redes neurais usando nnsight com execução remota NDIF opcional. Use quando precisar executar experimentos de interpretabilidade em modelos massivos (70B+) sem recursos locais de GPU, ou ao trabalhar com qualquer arquitetura PyTorch.

中文简介正在准备中。当前显示原文说明。

Fornece orientação para pesquisa de interpretabilidade mecanística usando TransformerLens para inspecionar e manipular internals de transformers via HookPoints e caching de ativações. Use ao fazer engenharia reversa de algoritmos de modelos, estudar padrões de atenção ou realizar experimentos de activation patching.

中文简介正在准备中。当前显示原文说明。

Provides guidance for interpreting and manipulating neural network internals using nnsight with optional NDIF remote execution. Use when needing to run interpretability experiments on massive models (70B+) without local GPU resources, or when working with any PyTorch architecture.

中文简介正在准备中。当前显示原文说明。

Use SAELens to train and analyze sparse autoencoders, which separate neural network activity into features that may represent recognizable concepts. Explore pretrained features, evaluate reconstruction quality, and investigate how individual features influence language model outputs.

中文简介正在准备中。当前显示英语简介。

• Exploring features with a pretrained SAE
• Training and evaluating a custom SAE
• Studying overlapping concepts in model activity

Use SAELens to train and analyze sparse autoencoders, which break model activations into a small set of interpretable features. Explore pretrained features, evaluate reconstruction quality, troubleshoot training, and test how individual features influence generated text.

中文简介正在准备中。当前显示英语简介。

• Exploring features in a pretrained SAE
• Training an SAE for a selected model layer
• Investigating inactive features and reconstruction errors

Use SAELens to train and analyze sparse autoencoders, which break model activations into a small set of interpretable features. Explore pretrained features, evaluate reconstruction quality, troubleshoot training, and test how individual features influence generated text.

中文简介正在准备中。当前显示英语简介。

• Exploring features in a pretrained SAE
• Training an SAE for a selected model layer
• Investigating inactive features and reconstruction errors

Designs retrospective or prospective clinical cohort study protocols for biomedical and clinical research. Always use this skill when the user needs a cohort-based study plan rather than a general study idea, evidence summary, or mechanistic experiment design. Focus on cohort appropriateness, enrollment logic, baseline time-zero definition, follow-up structure, endpoint definition, variable collection, confounding control, and a coherent primary statistical analysis line. Do not invent data availability, follow-up completeness, outcome ascertainment quality, sample size adequacy, or causal interpretability.

中文简介正在准备中。当前显示原文说明。

Fornece orientação para treinar e analisar Autoencodificadores Esparsos (SAEs) usando SAELens para decompor ativações de redes neurais em features interpretáveis. Use ao descobrir features interpretáveis, analisar superposição ou estudar representações monossemânticas em modelos de linguagem.

中文简介正在准备中。当前显示原文说明。

Provides guidance for training and analyzing Sparse Autoencoders (SAEs) using SAELens to decompose neural network activations into interpretable features. Use when discovering interpretable features, analyzing superposition, or studying monosemantic representations in language models.

中文简介正在准备中。当前显示原文说明。

Remove refusal behaviors from open-weight LLMs using OBLITERATUS — mechanistic interpretability techniques (diff-in-means, SVD, whitened SVD, LEACE, SAE decomposition, etc.) to excise guardrails while preserving reasoning. 9 CLI methods, 28 analysis modules, 116 model presets across 5 compute tiers, tournament evaluation, and telemetry-driven recommendations. Use when a user wants to uncensor, abliterate, or remove refusal from an LLM.

中文简介正在准备中。当前显示原文说明。

Provides guidance for training and analyzing Sparse Autoencoders (SAEs) using SAELens to decompose neural network activations into interpretable features. Use when discovering interpretable features, analyzing superposition, or studying monosemantic representations in language models.

中文简介正在准备中。当前显示原文说明。