INDEX

Explanations

phrases related to criticism or judgment, especially focused on labeling things as "stupid."

oai_token-act-pair · gpt-3.5-turbo

references to the concept of stupidity

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 8-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.8.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.8.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

AUT

-0.80

riott

-0.77

APH

-0.76

ILA

-0.73

apers

-0.73

accompan

-0.73

Reviewed

-0.72

orthy

-0.71

chwitz

-0.69

avez

-0.67

POSITIVE LOGITS

founded

1.09

found

0.96

nesses

0.92

ness

0.86

est

0.86

ly

0.84

fuck

0.83

itude

0.83

 shit

0.81

asses

0.80

Activations Density 0.047%

phrases related to criticism or judgment, especially focused on labeling things as "stupid."

references to the concept of stupidity

No Comments

No Known Activations