INDEX

Explanations

keywords related to clever or deceptive actions

oai_token-act-pair · gpt-3.5-turbo

references to deceptive or tricky behavior

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 7-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.7.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.7.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

 isot

-0.70

ccording

-0.70

ŃĶ

-0.70

ankind

-0.64

 Hemp

-0.63

 enrichment

-0.63

Interstitial

-0.62

 Expend

-0.62

large

-0.60

 Domain

-0.60

POSITIVE LOGITS

ery

1.30

ster

1.16

 tricks

1.02

sters

1.01

 trick

0.96

eries

0.96

aroo

0.94

 Trick

0.92

iest

0.91

door

0.91

Activations Density 0.026%

keywords related to clever or deceptive actions

references to deceptive or tricky behavior

No Comments

No Known Activations