INDEX

Explanations

the word "Thomson."

oai_token-act-pair · gpt-3.5-turbo

mentions of the entity "Thomson Reuters."

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 4-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.4.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.4.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

mble

-0.73

eer

-0.72

hov

-0.70

aeper

-0.69

OOL

-0.67

yg

-0.64

eers

-0.64

ee

-0.63

stract

-0.63

âĸ¬

-0.63

POSITIVE LOGITS

 Thomson

1.15

psons

1.01

 Reuters

0.96

DragonMagazine

0.88

Thom

0.86

sonian

0.85

ciating

0.85

stown

0.84

 Browne

0.83

iola

0.77

Activations Density 0.016%

the word "Thomson."

mentions of the entity "Thomson Reuters."

No Comments

No Known Activations