INDEX

Explanations

mentions of politicians and political activities

oai_token-act-pair · gpt-3.5-turbo Triggered by @bot

references to members of Congress

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 12-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.11.hook_resid_post

Prompts (Dashboard)

24,576 prompts, 128 tokens each

Dataset (Dashboard)

Skylion007/openwebtext

Features

24,576

Data Type

torch.float32

Hook Point

blocks.11.hook_resid_post

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

 underpin

-0.61

 pron

-0.58

 mistaken

-0.57

 fencing

-0.57

bul

-0.56

 AMERICA

-0.56

 mist

-0.55

 split

-0.55

 halves

-0.55

 pred

-0.54

POSITIVE LOGITS

orters

1.42

orter

1.23

rint

1.15

resents

1.10

utation

1.09

atri

1.06

rodu

1.06

orted

1.03

ository

1.03

airs

1.00

Activations Density 0.036%

mentions of politicians and political activities

references to members of Congress

No Comments

No Known Activations