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

Cu

-0.69

sav

-0.67

rgb

-0.67

hod

-0.63

resh

-0.63

 scram

-0.61

Zot

-0.61

factor

-0.61

untu

-0.60

rums

-0.60

POSITIVE LOGITS

Interstitial

0.68

 Rebellion

0.65

bidden

0.64

 dissatisf

0.64

ukemia

0.62

 folly

0.62

acca

0.59

lein

0.59

onis

0.58

bad

0.58

Activations Density 0.000%

No Known Activations

This feature has no known activations.