All authors
Claude Skills by VectorSpaceLab
github.com/VectorSpaceLab6,028 skills13 installs7,823 views
- Intervention Comparison MatrixCompare robustness interventions across clean and shifted metrics using direction-aware evidence signs.Votes: 0GitHub stars: 247
- Reduced Recovery EvaluatorAssemble executable soft-mode proxy recovery results with metrics, traces, and mechanism checks.Votes: 0GitHub stars: 247
- Robustness Benchmark ProtocolValidate clean and shifted robustness benchmark protocols with class-overlap and metric-gap contracts.Votes: 0GitHub stars: 247
- Detection Metric ProtocolCompute AUROC and AUPR detection metrics with paper-faithful score orientations for softmax confidence baselines.Votes: 0GitHub stars: 247
- Proxy Recovery HarnessRun a bounded mechanism-faithful proxy experiment for maximum softmax OOD detection recovery artifacts.Votes: 0GitHub stars: 247
- Softmax Confidence ScoringCompute maximum softmax probability detector scores from logits or probabilities for misclassification and OOD detection.Votes: 0GitHub stars: 247
- Collocation Morphology PreprocessorPreprocess English text with WordNet-style collocation matching, tokenization, and inflectional normalization.Votes: 0GitHub stars: 247
- Context Sense TaggerSelect or defer WordNet-style sense pointers using current and previous sentence context.Votes: 0GitHub stars: 247
- Lexical Taxonomy SchemaBuild and validate compact WordNet-style synset and semantic-pointer taxonomies for recovery experiments.Votes: 0GitHub stars: 247
- Semantic Distance EvaluatorEstimate WordNet-style semantic distance by traversing typed semantic-pointer graphs.Votes: 0GitHub stars: 247
- Taxonomy Recovery HarnessRun a reduced executable WordNet taxonomy recovery using generated preprocessing, tagging, and distance skills.Votes: 0GitHub stars: 247
- Dpo Loss ObjectiveCompute Direct Preference Optimization logistic losses, implicit rewards, and log-ratio diagnostics from policy and reference log probabilities.Votes: 0GitHub stars: 247
- Dpo Preference Data ContractNormalize pairwise preference records into prompt, chosen, rejected, and metadata fields for Direct Preference Optimization training.Votes: 0GitHub stars: 247
- Dpo Reduced Training HarnessRun a bounded scalar DPO optimization that records loss, parameter changes, preference accuracy, and mechanism checks for reduced recovery.Votes: 0GitHub stars: 247
- Dpo Sequence Logprob AccountingCompute response-only sequence log probabilities for DPO by masking prompts and padding before summing token log probabilities.Votes: 0GitHub stars: 247
- Ffn Activation Logit AttributionMeasure whether activated FFN neurons preferentially raise logits for their projected promoted tokens versus controls.Votes: 0GitHub stars: 247
- Ffn Concept GroupingGroup top promoted vocabulary tokens into human-readable concept labels with purity and unresolved-token evidence.Votes: 0GitHub stars: 247
- Ffn Proxy Recovery HarnessRun a bounded mechanism-faithful proxy experiment for FFN value-vector concept promotion using generated skills.Votes: 0GitHub stars: 247
- Ffn Value Vector ExtractionExtract and normalize feed-forward output value vectors from transformer-like model weights for concept-promotion analysis.Votes: 0GitHub stars: 247
- Ffn Vocabulary ProjectionProject FFN value vectors through an LM head to rank promoted tokens for mechanistic concept analysis.Votes: 0GitHub stars: 247
- Pplm Attribute ObjectivesCompute PPLM-style bag-of-words or linear-classifier attribute losses and gradients for controlled generation.Votes: 0GitHub stars: 247
- Pplm Controlled Generation EvaluationEvaluate PPLM controlled-generation proxies with target-mass gain, KL fluency cost, and target consistency checks.Votes: 0GitHub stars: 247
- Pplm Fusion GenerationFuse PPLM perturbed and unperturbed token distributions using geometric mixing for fluent controlled decoding.Votes: 0GitHub stars: 247
- Pplm Perturbation LoopRun PPLM-style iterative normalized gradient perturbations with KL regularization and optimizer trace evidence.Votes: 0GitHub stars: 247
- Rtp Continuation Generation ProtocolCreate bounded multi-continuation records per prompt for toxic degeneration evaluation with explicit generator metadata.Votes: 0GitHub stars: 247
- Rtp Prompt Dataset ProtocolNormalize RealToxicityPrompts-style prompt records for toxic and non-toxic prompted generation evaluation.Votes: 0GitHub stars: 247
- Rtp Recovery Experiment HarnessRun an executable bounded recovery harness that composes prompt normalization, generation, scoring, and aggregation with mechanism checks.Votes: 0GitHub stars: 247
- Rtp Toxicity Metric AggregationCompute expected maximum toxicity and toxicity probability for RealToxicityPrompts-style continuation sets.Votes: 0GitHub stars: 247
- Rtp Toxicity Scoring AdapterAttach numeric toxicity scores to generated continuations using Perspective API or declared offline proxy scoring.Votes: 0GitHub stars: 247
- Attention Path ExpansionCompute frozen-attention direct, head, and virtual-head path contributions for attention-only circuits.Votes: 0GitHub stars: 247
- Circuit Recovery HarnessBuild and validate bounded mechanism-faithful recovery artifacts for Transformer Circuits proxy experiments.Votes: 0GitHub stars: 247
- Induction Head DetectorDetect induction-head copying behavior on repeated token sequences with explicit mechanism checks.Votes: 0GitHub stars: 247
- Qk Ov Circuit ExpansionExpand attention-head QK and OV weights into token-level circuit matrices and copying diagnostics.Votes: 0GitHub stars: 247
- Residual Logit LensApply logit-lens unembedding and additive residual contribution checks for mechanistic transformer circuit analysis.Votes: 0GitHub stars: 247
- Curriculum RegularizationGenerate coefficient schedules for curriculum-regularized PINN recovery experiments.Votes: 0GitHub stars: 247
- Failure Mode DiagnosticsDiagnose PINN failure-mode recovery runs with relative errors and optimizer progress checks.Votes: 0GitHub stars: 247
- Periodic Pde BenchmarkBuild deterministic periodic PDE benchmarks for PINN failure-mode recovery experiments.Votes: 0GitHub stars: 247
- Pinn Residual ObjectiveCompute PINN data, boundary, and PDE residual losses for reduced failure-mode experiments.Votes: 0GitHub stars: 247
- Reduced Recovery HarnessRun bounded reduced PINN recovery experiments that exercise generated benchmark, objective, diagnostics, and curriculum skills.Votes: 0GitHub stars: 247
- Deep Ritz Residual Trial NetworkBuild smooth residual trial networks for Deep Ritz variational PDE objectives.Votes: 0GitHub stars: 247
- Deep Ritz Stochastic Quadrature SamplingGenerate stochastic interior and boundary quadrature samples for Deep Ritz variational PDE training.Votes: 0GitHub stars: 247
- Deep Ritz Training RecoveryRun bounded Deep Ritz optimization and emit auditable recovery artifacts for variational PDE experiments.Votes: 0GitHub stars: 247
- Deep Ritz Variational Energy LossCompute Deep Ritz variational energy losses with gradient energy and boundary penalties.Votes: 0GitHub stars: 247
- Gated Pinn ArchitectureBuild a lightweight gated PINN-style scalar model that exposes trainable parameters and gradient paths for reduced recovery experiments.Votes: 0GitHub stars: 247
- Gradient Stat AnnealingUpdate PINN data-loss weights from residual and data-fit gradient statistics using the paper's moving-average annealing rule.Votes: 0GitHub stars: 247
- Helmholtz Pinn ProblemConstruct deterministic reduced Helmholtz PINN benchmark data with analytic solution, boundary samples, collocation samples, forcing, and relative L2 scoring.Votes: 0GitHub stars: 247
- Pinn Loss DecompositionCompute separated Helmholtz PINN residual and boundary losses so gradient imbalance can be diagnosed and corrected.Votes: 0GitHub stars: 247
- Reduced Recovery EvaluationRun bounded reduced Helmholtz PINN recovery with optimizer updates, lambda traces, relative L2 metrics, and validator-compatible evidence.Votes: 0GitHub stars: 247
- Curvature MemoryMaintain bounded positive-curvature L-BFGS correction memory for large-scale quasi-Newton optimization.Votes: 0GitHub stars: 247
- Proxy Recovery EvaluationBuild validator-compatible soft-mode recovery metrics and mechanism checks for L-BFGS proxy experiments.Votes: 0GitHub stars: 247