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

teness

-0.63

amination

-0.62

long

-0.61

ãĤ¢

-0.59

abouts

-0.59

avis

-0.58

dding

-0.58

 Codec

-0.57

iod

-0.56

 Doyle

-0.56

POSITIVE LOGITS

iencies

0.90

amsung

0.82

erves

0.74

jri

0.73

 grapes

0.73

 Rothschild

0.73

ushi

0.71

enegger

0.71

hiba

0.68

ashtra

0.68

Activations Density 0.000%

No Known Activations

This feature has no known activations.