INDEX

Explanations

phrases indicating support or approval

oai_token-act-pair · gpt-3.5-turbo

phrases related to support or endorsement of various policies or causes

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 8-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.8.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.8.hook_resid_pre

Architecture

standard

Context Size

128

Dataset

Skylion007/openwebtext

Hook Point Layer

Activation Function

relu

Embeds

IFrame

Link

Not in Any Lists

No Comments

Negative Logits

DragonMagazine

-0.80

Â¯Â¯Â¯Â¯Â¯Â¯Â¯Â¯Â¯Â¯Â¯Â¯Â¯Â¯Â¯Â¯

-0.74

ngth

-0.69

Buzz

-0.65

opard

-0.64

semble

-0.63

VI

-0.63

ertation

-0.63

 stumble

-0.62

hemer

-0.61

POSITIVE LOGITS

 abol

1.22

 decriminal

1.15

 legalizing

1.12

 banning

1.07

 repeal

1.03

 stricter

1.03

 abolition

1.01

 inclusion

1.01

 legalization

1.00

 repealing

0.99

Activations Density 0.299%

phrases indicating support or approval

phrases related to support or endorsement of various policies or causes

No Comments

No Known Activations