INDEX

Explanations

names or terms starting with "Be"

oai_token-act-pair · gpt-3.5-turbo

the word "Be" and its various forms or variations throughout the text

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 11-res-jb

Configuration

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

 pole

-0.67

 mash

-0.66

 breeze

-0.64

 Crunch

-0.63

 escalation

-0.61

 drift

-0.61

 glide

-0.61

 succeeding

-0.60

 dispatch

-0.60

 being

-0.59

POSITIVE LOGITS

aum

1.37

arer

1.29

yond

1.22

hemoth

1.22

arded

1.21

arers

1.21

asley

1.21

acons

1.16

auc

1.14

cca

1.13

Activations Density 0.025%

names or terms starting with "Be"

the word "Be" and its various forms or variations throughout the text

No Comments

No Known Activations