INDEX

Explanations

phrases related to security and access control procedures

oai_token-act-pair · gpt-3.5-turbo Triggered by @bot

repetitive patterns and references to absence or voids

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 12-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.11.hook_resid_post

Prompts (Dashboard)

24,576 prompts, 128 tokens each

Dataset (Dashboard)

Skylion007/openwebtext

Features

24,576

Data Type

torch.float32

Hook Point

blocks.11.hook_resid_post

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

eele

-0.81

artney

-0.74

ruck

-0.71

EStream

-0.66

interstitial

-0.65

ãĤ¼ãĤ¦ãĤ¹

-0.64

avis

-0.64

veland

-0.63

abre

-0.62

etch

-0.61

POSITIVE LOGITS

no

1.56

no

1.40

 nothing

1.24

NO

1.23

No

1.23

No

1.19

 nobody

1.15

 none

1.15

nothing

1.09

 zero

1.07

Activations Density 0.353%

phrases related to security and access control procedures

repetitive patterns and references to absence or voids

No Comments

No Known Activations