INDEX

Explanations

the word "does" in different contexts

oai_token-act-pair · gpt-3.5-turbo

instances of the word "does."

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 4-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.4.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.4.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

 Canal

-0.73

 Unle

-0.72

 Typhoon

-0.70

ESA

-0.70

 palms

-0.68

 Ascension

-0.67

 Methods

-0.66

Tai

-0.65

haul

-0.63

 Vers

-0.63

POSITIVE LOGITS

omething

1.06

ppel

0.88

not

0.86

berra

0.81

ettings

0.81

ĸļ

0.80

paces

0.78

NOT

0.78

omorphic

0.78

 confir

0.76

Activations Density 0.041%

the word "does" in different contexts

instances of the word "does."

No Comments

No Known Activations