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

jc

-0.80

ouf

-0.77

ynamic

-0.77

@@

-0.72

rums

-0.67

tenance

-0.66

nc

-0.62

TION

-0.61

haps

-0.61

kson

-0.60

POSITIVE LOGITS

 tremend

0.80

 occas

0.71

 occasion

0.71

avorite

0.70

 persecut

0.66

å§«

0.64

 Paragu

0.64

Laughs

0.62

MpServer

0.62

track

0.61

Activations Density 0.000%

No Known Activations

This feature has no known activations.