INDEX

Explanations

locations, organizations, and names related to military and government operations

oai_token-act-pair · gpt-3.5-turbo

references to specific government agencies and organizations involved in surveillance and defense

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 4-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.4.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.4.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

¥µ

-0.74

utics

-0.67

 Roth

-0.61

 Huck

-0.61

 Blizz

-0.60

avor

-0.60

BU

-0.59

ending

-0.59

ulf

-0.59

hesis

-0.58

POSITIVE LOGITS

CHQ

1.14

raltar

0.95

agnetic

0.87

illin

0.84

æ©Ł

0.80

obyl

0.75

å§«

0.74

nown

0.73

emouth

0.72

anyon

0.72

Activations Density 0.014%

locations, organizations, and names related to military and government operations

references to specific government agencies and organizations involved in surveillance and defense

No Comments

No Known Activations