INDEX

Explanations

the word "fellow" followed by another word

oai_token-act-pair · gpt-3.5-turbo

references to acquaintances or peers

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

ARA

-0.73

anish

-0.71

anwhile

-0.70

 seiz

-0.70

 Retrieved

-0.67

 Ukrain

-0.65

ADE

-0.63

onies

-0.63

asive

-0.63

 DRAG

-0.62

POSITIVE LOGITS

 classmates

0.87

 traveler

0.86

 travellers

0.86

 travelers

0.84

 strugg

0.83

 fellow

0.83

workers

0.81

 traveller

0.79

member

0.79

worldly

0.78

Activations Density 0.009%

the word "fellow" followed by another word

references to acquaintances or peers

No Comments

No Known Activations