INDEX

Explanations

the word "nothing" followed by high activations

oai_token-act-pair · gpt-3.5-turbo

negative assertions and phrases emphasizing nullity or insignificance

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 7-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.7.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.7.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

PLA

-0.67

landers

-0.63

uctions

-0.59

eus

-0.59

 Bots

-0.58

 transitions

-0.58

 decline

-0.57

eton

-0.57

 prohibitions

-0.57

 downs

-0.57

POSITIVE LOGITS

lled

0.86

avering

0.75

bered

0.74

umbn

0.73

arily

0.73

ient

0.72

ĸļ

0.69

 akin

0.69

itter

0.68

ozyg

0.68

Activations Density 0.100%

the word "nothing" followed by high activations

negative assertions and phrases emphasizing nullity or insignificance

No Comments

No Known Activations