INDEX

Explanations

No Explanations Found

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 0-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.0.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.0.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

Ħ¢

-0.88

 teasp

-0.87

 nodd

-0.85

 tiss

-0.81

©¶æ¥µ

-0.81

 gobl

-0.80

ļéĨĴ

-0.78

srf

-0.75

bilt

-0.74

arbon

-0.74

POSITIVE LOGITS

uras

0.66

apego

0.66

fact

0.65

yet

0.65

itizen

0.65

 congress

0.65

actionDate

0.64

Story

0.64

iaries

0.64

iary

0.63

Activations Density 0.000%

No Known Activations

This feature has no known activations.