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

LOD

-0.83

nah

-0.82

externalActionCode

-0.80

ioxide

-0.75

doc

-0.73

sbm

-0.72

abama

-0.72

VERSION

-0.71

hum

-0.71

esm

-0.71

POSITIVE LOGITS

 targets

0.71

ãĥ¯

0.70

soever

0.70

 Cabrera

0.69

ĻĤ

0.65

ãĤ¹ãĥĪ

0.65

¢

0.65

ļ

0.64

ãĥ¥

0.61

 Rivera

0.61

Activations Density 0.000%

No Known Activations

This feature has no known activations.