INDEX

Explanations

phrases encouraging action or expression

oai_token-act-pair · gpt-3.5-turbo

phrases related to action and assertiveness in a political context

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 10-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.10.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.10.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

ļéĨĴ

-0.76

lav

-0.71

wic

-0.70

 Weekly

-0.63

 Intake

-0.63

riad

-0.61

 Testing

-0.60

Luck

-0.60

imore

-0.60

eties

-0.60

POSITIVE LOGITS

 declare

1.16

 proclaim

1.15

 admit

1.03

 scream

1.00

 shout

0.99

 embrace

0.96

say

0.95

 apologize

0.94

 announce

0.94

 unleash

0.94

Activations Density 0.130%

phrases encouraging action or expression

phrases related to action and assertiveness in a political context

No Comments

No Known Activations