INDEX

Explanations

phrases related to prestigious awards and honors

oai_token-act-pair · gpt-3.5-turbo

references to prestigious awards, particularly the Nobel Prize and Pulitzer Prize

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

icago

-0.78

aves

-0.72

yy

-0.69

addock

-0.68

 Antar

-0.67

alach

-0.67

andering

-0.66

oute

-0.65

ECTION

-0.63

mson

-0.63

POSITIVE LOGITS

 Prize

1.37

 laureate

1.18

 Nobel

1.03

 laure

1.01

 prize

0.94

 Laure

0.92

Nob

0.91

 prizes

0.80

Pri

0.80

 Peace

0.79

Activations Density 0.009%

phrases related to prestigious awards and honors

references to prestigious awards, particularly the Nobel Prize and Pulitzer Prize

No Comments

No Known Activations