INDEX

Explanations

phrases indicating the speaker has seen something, often emphasizing it

oai_token-act-pair · gpt-3.5-turbo

phrases indicating something that has been experienced or witnessed

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

ÃŃa

-0.78

 agric

-0.74

atto

-0.68

vernment

-0.68

trap

-0.67

externalActionCode

-0.67

 mantle

-0.64

 contrace

-0.64

currency

-0.64

vice

-0.63

POSITIVE LOGITS

dust

0.87

 firsthand

0.82

ewitness

0.78

Ĺ

0.78

finder

0.72

agame

0.71

 glimps

0.69

 footage

0.68

peak

0.68

 Seen

0.66

Activations Density 0.041%

phrases indicating the speaker has seen something, often emphasizing it

phrases indicating something that has been experienced or witnessed

No Comments

No Known Activations