INDEX

Explanations

phrases indicating something observed or happening up to a certain point in time

oai_token-act-pair · gpt-3.5-turbo

phrases that indicate progression or distance in time

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 8-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.8.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.8.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

-+

-0.65

cause

-0.65

kefeller

-0.65

andal

-0.64

equal

-0.62

MpServer

-0.61

me

-0.61

imb

-0.60

inges

-0.59

SourceFile

-0.59

POSITIVE LOGITS

 there

0.90

we

0.87

adays

0.78

 however

0.76

 nobody

0.75

 they

0.72

 though

0.69

 everyone

0.66

 everybody

0.65

it

0.63

Activations Density 0.113%

phrases indicating something observed or happening up to a certain point in time

phrases that indicate progression or distance in time

No Comments

No Known Activations