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

ensable

-0.73

inant

-0.69

obbies

-0.68

ylum

-0.65

ANE

-0.64

atre

-0.63

FTA

-0.62

raints

-0.61

 actionGroup

-0.60

ISION

-0.60

POSITIVE LOGITS

uther

0.66

 quotation

0.64

env

0.64

topic

0.62

json

0.62

ilar

0.61

 Petersen

0.61

"@

0.59

Cur

0.59

blown

0.59

Activations Density 0.000%

No Known Activations

This feature has no known activations.