INDEX

Explanations

keywords related to espionage and secret information

oai_token-act-pair · gpt-3.5-turbo

references to hidden or secret knowledge related to world events

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

 Weld

-0.70

 âī¡

-0.68

 Ross

-0.66

 Clyde

-0.63

 Irwin

-0.62

 Tenn

-0.62

 Ripple

-0.62

 Kathryn

-0.61

 Conflict

-0.60

 Kenny

-0.60

POSITIVE LOGITS

gallery

0.91

hillary

0.91

recomm

0.90

photos

0.90

dead

0.88

cond

0.88

terms

0.88

committee

0.87

cook

0.87

warm

0.87

Activations Density 0.074%

keywords related to espionage and secret information

references to hidden or secret knowledge related to world events

No Comments

No Known Activations