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

pron

-0.91

sea

-0.78

wr

-0.76

angular

-0.74

orb

-0.73

erg

-0.71

raped

-0.70

sex

-0.69

romeda

-0.69

{\

-0.68

POSITIVE LOGITS

phabet

0.69

 Duff

0.66

ixels

0.65

 hops

0.64

 Ammunition

0.63

Dek

0.62

 Charter

0.62

ioxide

0.62

 optimizations

0.62

 Â®

0.61

Activations Density 0.000%

No Known Activations

This feature has no known activations.