INDEX

Explanations

phrases indicating the addition of information or details

oai_token-act-pair · gpt-3.5-turbo

instances of the word "Add" likely indicating instructions or commands to include something

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 5-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.5.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.5.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

 Pulse

-0.69

creen

-0.68

 writ

-0.65

fal

-0.63

 unim

-0.61

 Maker

-0.61

Fal

-0.60

 Falcon

-0.60

PATH

-0.60

ZI

-0.60

POSITIVE LOGITS

endum

1.32

ition

1.31

icted

1.19

itions

1.16

ressing

1.16

resses

1.14

icts

1.14

ictions

1.01

ressed

0.99

itionally

0.97

Activations Density 0.020%

phrases indicating the addition of information or details

instances of the word "Add" likely indicating instructions or commands to include something

No Comments

No Known Activations