INDEX

Explanations

names or titles of executive positions

oai_token-act-pair · gpt-3.5-turbo

references to individuals with executive titles or roles

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

apo

-0.81

ERO

-0.80

bane

-0.76

ologies

-0.75

ulia

-0.74

abouts

-0.74

cale

-0.73

ILLE

-0.72

some

-0.72

lust

-0.72

POSITIVE LOGITS

 director

0.98

 producer

0.97

 summary

0.89

 committee

0.86

utive

0.85

 chef

0.84

 assistant

0.83

director

0.82

 officer

0.81

 chairman

0.81

Activations Density 0.020%

names or titles of executive positions

references to individuals with executive titles or roles

No Comments

No Known Activations