INDEX

Explanations

phrases related to protection or defense

oai_token-act-pair · gpt-3.5-turbo

phrases related to protecting or shielding from various dangers or threats

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 7-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.7.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.7.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

mun

-0.75

ftime

-0.73

SELECT

-0.72

page

-0.72

pring

-0.70

arse

-0.70

Switch

-0.69

ournals

-0.69

TEXT

-0.68

okes

-0.68

POSITIVE LOGITS

 afar

0.97

 harm

0.92

 tyranny

0.82

 extinction

0.82

 encro

0.81

 impending

0.76

 liability

0.76

 drowning

0.76

 pesky

0.76

 injury

0.75

Activations Density 0.062%

phrases related to protection or defense

phrases related to protecting or shielding from various dangers or threats

No Comments

No Known Activations