INDEX

Explanations

negative or violent descriptive words including physical harm or emotional distress

oai_token-act-pair · gpt-3.5-turbo

terms related to severe and morally reprehensible actions or events

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 6-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.6.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.6.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

arten

-0.83

essor

-0.82

pai

-0.82

inoa

-0.73

agall

-0.72

ership

-0.71

haw

-0.70

sers

-0.68

glas

-0.68

verning

-0.68

POSITIVE LOGITS

 atrocities

0.94

 horrors

0.90

 torture

0.82

 spectacle

0.80

 cruelty

0.80

 atroc

0.79

 murder

0.76

 asylum

0.76

 misogyn

0.75

 injustice

0.74

Activations Density 0.095%

negative or violent descriptive words including physical harm or emotional distress

terms related to severe and morally reprehensible actions or events

No Comments

No Known Activations