INDEX

Explanations

mentions of actions or discussions regarding interventions, especially in political or conflict contexts

oai_token-act-pair · gpt-3.5-turbo

instances of the word "intervention" and related terms

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

Export

-0.75

carbon

-0.74

cale

-0.73

WE

-0.73

Î»

-0.72

Boh

-0.70

\\\\\\\\

-0.65

ndra

-0.64

versions

-0.64

cake

-0.64

POSITIVE LOGITS

ateral

1.02

aries

0.97

anke

0.92

 interven

0.91

arie

0.87

 intervention

0.87

arial

0.86

ulatory

0.80

ist

0.79

 intervene

0.79

Activations Density 0.018%

mentions of actions or discussions regarding interventions, especially in political or conflict contexts

instances of the word "intervention" and related terms

No Comments

No Known Activations