INDEX

Explanations

the word "affer" with different variations and contexts

oai_token-act-pair · gpt-3.5-turbo

references to specific individuals or public figures

oai_token-act-pair · gpt-4o-mini Triggered by @bot

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

 supers

-0.66

Ho

-0.65

 stone

-0.64

Ore

-0.62

Ind

-0.62

 moons

-0.62

 token

-0.62

bye

-0.61

NS

-0.60

 together

-0.60

POSITIVE LOGITS

affer

4.81

antz

1.16

acus

1.03

aff

1.00

pherd

0.95

irlf

0.90

affe

0.88

apter

0.87

afer

0.87

uggage

0.87

Activations Density 0.016%

the word "affer" with different variations and contexts

references to specific individuals or public figures

No Comments

No Known Activations