INDEX

Explanations

references to suspicious or suspect activity

oai_token-act-pair · gpt-3.5-turbo

terms related to suspicion or being suspicious

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

eer

-0.73

 Borough

-0.70

ORGE

-0.69

 trader

-0.64

 grievance

-0.64

 professionally

-0.63

 baker

-0.63

 learning

-0.63

 living

-0.62

 Telephone

-0.62

POSITIVE LOGITS

ension

1.51

ensions

1.46

icious

1.31

ended

1.29

ending

1.10

endant

1.05

ention

1.03

enders

1.02

aces

1.02

heric

1.00

Activations Density 0.020%

references to suspicious or suspect activity

terms related to suspicion or being suspicious

No Comments

No Known Activations