INDEX

Explanations

text related to suspicion or suspicion-related terms

oai_token-act-pair · gpt-3.5-turbo

terms related to "suspension" or "suspicious activities."

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 3-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.3.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.3.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

eer

-0.73

EngineDebug

-0.65

SpaceEngineers

-0.65

 Borough

-0.65

 congr

-0.62

 guiActiveUnfocused

-0.62

ļéĨĴ

-0.62

 Cour

-0.61

 trib

-0.60

gie

-0.59

POSITIVE LOGITS

ensions

1.31

ension

1.30

icious

1.23

usp

1.11

ended

1.09

ugal

1.01

olicy

0.99

endant

0.98

enses

0.95

ention

0.94

Activations Density 0.021%

text related to suspicion or suspicion-related terms

terms related to "suspension" or "suspicious activities."

No Comments

No Known Activations