INDEX

Explanations

No Explanations Found

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 3-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.3.hook_resid_pre

Prompts (Dashboard)

24,576 prompts, 128 tokens each

Dataset (Dashboard)

Skylion007/openwebtext

Features

24,576

Data Type

torch.float32

Hook Point

blocks.3.hook_resid_pre

Architecture

standard

Context Size

128

Dataset

Skylion007/openwebtext

Hook Point Layer

Activation Function

relu

Embeds

PlotsExplanationShow Test FieldDefault Test Text

IFrame

Link

Not in Any Lists

No Comments

Negative Logits

pmwiki

-0.85

SourceFile

-0.79

ï¸ı

-0.78

DragonMagazine

-0.76

staff

-0.74

heit

-0.73

uploads

-0.69

oho

-0.69

SpaceEngineers

-0.68

":"","

-0.67

POSITIVE LOGITS

 derail

0.75

 conclude

0.66

 blender

0.62

 labels

0.61

 hypotheses

0.61

 happen

0.60

 mars

0.60

 derailed

0.60

 ramps

0.60

igators

0.59

Activations Density 0.000%

No Known Activations

This feature has no known activations.