INDEX

Explanations

references to certain people or entities, possibly related to specific topics or industries

oai_token-act-pair · gpt-3.5-turbo

references to the concept of "opinion."

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

ãĤ¼ãĤ¦ãĤ¹

-0.86

wagen

-0.77

åŃĲ

-0.76

ILLE

-0.70

ãĤ£

-0.69

 wart

-0.68

¶ħ

-0.68

=~=~

-0.66

 comprom

-0.64

éĩ

-0.63

POSITIVE LOGITS

aque

1.12

inion

1.10

yright

1.07

osite

1.07

aqu

1.06

onent

1.03

iate

0.99

codes

0.97

iates

0.97

rah

0.95

Activations Density 0.025%

references to certain people or entities, possibly related to specific topics or industries

references to the concept of "opinion."

No Comments

No Known Activations