INDEX

Explanations

phrases indicating physical or emotional discomfort

oai_token-act-pair · gpt-3.5-turbo

instances of the word "in" repeated frequently throughout the document

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

LER

-0.70

heast

-0.67

isks

-0.66

 Secondly

-0.62

 aside

-0.62

 Architects

-0.61

deal

-0.59

trop

-0.59

 WITHOUT

-0.58

[/

-0.58

POSITIVE LOGITS

 disbelief

1.33

 amaz

1.29

 agony

1.28

 aston

1.27

 frustration

1.24

 delight

1.22

 anguish

1.21

 ecstasy

1.20

 excitement

1.19

 laughter

1.18

Activations Density 0.120%

phrases indicating physical or emotional discomfort

instances of the word "in" repeated frequently throughout the document

No Comments

No Known Activations