INDEX

Explanations

phrases related to causing harm or damage

oai_token-act-pair · gpt-3.5-turbo

expressions of disruption or harm caused by actions or comments

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

cius

-0.84

 Quartz

-0.82

chrom

-0.71

regulated

-0.70

antz

-0.66

 Koen

-0.65

QU

-0.64

oglobin

-0.64

 Processing

-0.63

 Classification

-0.63

POSITIVE LOGITS

 havoc

1.44

 carnage

0.90

 mayhem

0.89

 damage

0.86

 devastation

0.84

 inflicted

0.82

 wounds

0.82

wre

0.81

efully

0.80

 untold

0.80

Activations Density 0.082%

phrases related to causing harm or damage

expressions of disruption or harm caused by actions or comments

No Comments

No Known Activations