INDEX

Explanations

mentions of the word "friend"

oai_token-act-pair · gpt-3.5-turbo

references to friends or relationships

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

 stage

-0.75

rot

-0.74

 crystall

-0.68

 minimum

-0.67

 continuous

-0.65

 hydraulic

-0.63

 stretch

-0.63

 secondary

-0.62

 processed

-0.62

 ammunition

-0.61

POSITIVE LOGITS

friend

4.59

friends

2.64

Friend

2.18

riend

1.98

 Friend

1.40

Friends

1.34

brother

1.25

friendly

1.25

love

1.17

 friend

1.11

Activations Density 0.024%

mentions of the word "friend"

references to friends or relationships

No Comments

No Known Activations