INDEX

Explanations

the word "ve" with differing activation values, potentially related to app optimization or rentals

oai_token-act-pair · gpt-3.5-turbo

the presence of the suffix "ve" in words

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 5-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.5.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.5.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

£ı

-0.78

olicy

-0.75

 GOODMAN

-0.72

assian

-0.69

matically

-0.67

artifacts

-0.66

 administ

-0.65

omo

-0.65

wcs

-0.62

anmar

-0.62

POSITIVE LOGITS

illance

1.16

mber

1.13

ttes

1.02

ggie

0.98

ller

0.97

llers

0.95

lla

0.94

tsy

0.94

tted

0.94

tt

0.94

Activations Density 0.037%

the word "ve" with differing activation values, potentially related to app optimization or rentals

the presence of the suffix "ve" in words

No Comments

No Known Activations