INDEX

Explanations

instances of the word "Executive" followed by a single-digit number

oai_token-act-pair · gpt-3.5-turbo

references to executive roles and positions

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 0-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.0.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.0.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

 scrape

-0.74

 virginity

-0.70

 rout

-0.67

 fend

-0.66

ques

-0.65

ppa

-0.65

 swarm

-0.64

 clicked

-0.64

 dunk

-0.64

opes

-0.64

POSITIVE LOGITS

 Executive

3.71

Executive

2.60

 executive

2.56

 Exec

1.44

exec

1.35

 exec

1.28

 Legislative

1.26

 Administration

1.23

 Judicial

1.23

 Presidential

1.21

Activations Density 0.013%

instances of the word "Executive" followed by a single-digit number

references to executive roles and positions

No Comments

No Known Activations