INDEX

Explanations

No Explanations Found

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 3-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.3.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.3.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

 Tales

-0.65

 slic

-0.64

 claimant

-0.64

 composing

-0.64

 refinement

-0.63

MX

-0.63

 Printed

-0.62

 ancestor

-0.62

rome

-0.62

 ancestors

-0.58

POSITIVE LOGITS

jad

0.81

ablishment

0.81

afia

0.80

obin

0.75

chieve

0.75

urga

0.74

 Canaver

0.72

ileaks

0.71

 Debor

0.71

isson

0.71

Activations Density 0.000%

No Known Activations

This feature has no known activations.