INDEX

Explanations

references to close relationships or organizations related to companionship

oai_token-act-pair · gpt-3.5-turbo

mentions of the word "Friends" in various contexts

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 0-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.0.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.0.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

 churn

-0.72

 belt

-0.68

 cylinder

-0.67

 shedding

-0.66

 tipped

-0.65

dep

-0.64

 vein

-0.62

AMD

-0.62

ossus

-0.62

 saturation

-0.60

POSITIVE LOGITS

 Friends

4.02

Friends

2.95

friends

2.34

 Friend

1.82

 friends

1.81

 Friendship

1.63

FRI

1.61

friend

1.59

 pals

1.47

Friend

1.44

Activations Density 0.016%

references to close relationships or organizations related to companionship

mentions of the word "Friends" in various contexts

No Comments

No Known Activations