INDEX

Explanations

terms related to authority figures in a workplace setting

oai_token-act-pair · gpt-3.5-turbo

references to supervisory roles or positions

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 5-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.5.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.5.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

ted

-0.79

rams

-0.78

bed

-0.76

çīĪ

-0.74

glers

-0.71

rums

-0.71

aciously

-0.71

oliath

-0.70

elin

-0.70

auntlets

-0.70

POSITIVE LOGITS

 supervisor

1.01

 Supervisor

0.86

 supervisors

0.83

NBC

0.75

 Sergeant

0.73

 ANGEL

0.70

intendent

0.69

 overseeing

0.68

terday

0.68

plin

0.68

Activations Density 0.017%

terms related to authority figures in a workplace setting

references to supervisory roles or positions

No Comments

No Known Activations