INDEX

Explanations

No Explanations Found

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 1-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.1.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.1.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

matic

-0.82

sonian

-0.77

furt

-0.76

 âĨĳ

-0.75

 âĶľ

-0.74

APD

-0.73

 showc

-0.72

MAT

-0.72

oÄŁ

-0.70

ãĥ¯ãĥ³

-0.70

POSITIVE LOGITS

 suppose

0.70

 somehow

0.65

 cloaked

0.64

abad

0.62

 Allaah

0.62

 definitely

0.61

 dearly

0.61

 surely

0.61

 detectable

0.61

 starship

0.61

Activations Density 0.000%

No Known Activations

This feature has no known activations.