INDEX

Explanations

mentions of comedy-related words or phrases

oai_token-act-pair · gpt-3.5-turbo

references to comedy and humor

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

IFrame

Link

Not in Any Lists

No Comments

Negative Logits

igree

-0.76

eded

-0.75

ipped

-0.73

hips

-0.69

ignty

-0.67

ainer

-0.67

irts

-0.66

rain

-0.65

installed

-0.64

ims

-0.64

POSITIVE LOGITS

 comedy

0.99

 comed

0.92

 Comed

0.88

 theatre

0.88

 sketches

0.86

 theater

0.86

 comedian

0.84

 Schumer

0.84

 comedians

0.83

 Comedy

0.82

Activations Density 0.015%

mentions of comedy-related words or phrases

references to comedy and humor

No Comments

No Known Activations