INDEX

Explanations

phrases indicating a request or call for action

oai_token-act-pair · gpt-3.5-turbo

phrases that indicate calls to action or requests directed at individuals or authorities

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 9-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.9.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.9.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

Kop

-0.75

 Malk

-0.74

anka

-0.61

Flo

-0.61

icka

-0.61

traumatic

-0.60

 cheered

-0.59

ults

-0.59

teen

-0.58

growing

-0.58

POSITIVE LOGITS

 reconsider

1.48

 intervene

1.29

 clarify

1.26

 postpone

1.21

 disclose

1.20

 refrain

1.20

 adopt

1.20

 cease

1.18

 rethink

1.17

 abandon

1.16

Activations Density 0.118%

phrases indicating a request or call for action

phrases that indicate calls to action or requests directed at individuals or authorities

No Comments

No Known Activations