INDEX

Explanations

phrases related to conspiracy and collusion

oai_token-act-pair · gpt-3.5-turbo

references to collusion and conspiracy

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

Hop

-0.80

aghd

-0.71

adish

-0.70

fly

-0.67

care

-0.67

ulton

-0.67

aign

-0.66

mac

-0.66

 Derby

-0.66

rowth

-0.65

POSITIVE LOGITS

 collusion

1.02

 complicit

0.97

 complicity

0.92

 conspiring

0.92

 unfocusedRange

0.89

 wink

0.82

atorial

0.81

 deceived

0.72

icate

0.71

eering

0.70

Activations Density 0.024%

phrases related to conspiracy and collusion

references to collusion and conspiracy

No Comments

No Known Activations