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

gae

-0.85

iliar

-0.80

locking

-0.78

Ļ½

-0.77

sels

-0.77

anders

-0.76

ealing

-0.74

aldi

-0.74

ernels

-0.71

agi

-0.69

POSITIVE LOGITS

WF

0.70

\\\\

0.66

ariat

0.64

EP

0.64

EQ

0.64

 shoulders

0.63

PW

0.62

 GOODMAN

0.62

UP

0.62

IU

0.62

Activations Density 0.000%

No Known Activations

This feature has no known activations.