INDEX

Explanations

phrases related to communication and giving information or instructions

oai_token-act-pair · gpt-3.5-turbo

instances of the word "tell" and related phrases indicating communication or questioning

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 10-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.10.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.10.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

grim

-0.66

apo

-0.63

ilogy

-0.62

ignt

-0.62

utton

-0.60

©¶æ¥µ

-0.60

VIDEO

-0.59

hra

-0.59

 Returning

-0.59

ELS

-0.57

POSITIVE LOGITS

 yourselves

2.00

 yourself

1.87

me

1.36

 Yourself

1.32

 your

1.24

us

1.04

 YOUR

0.98

ME

0.92

 somebody

0.80

your

0.80

Activations Density 0.596%

phrases related to communication and giving information or instructions

instances of the word "tell" and related phrases indicating communication or questioning

No Comments

No Known Activations