INDEX

Explanations

instances of actions or statements that could be seen as harmful or disruptive

oai_token-act-pair · gpt-3.5-turbo

phrases that convey action or inquiry

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 9-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.9.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.9.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

nces

-0.75

abouts

-0.71

sequent

-0.71

etheless

-0.70

webkit

-0.68

Keys

-0.68

 thereafter

-0.66

tions

-0.66

iann

-0.65

none

-0.62

POSITIVE LOGITS

 wrong

0.93

 Wrong

0.84

 delusional

0.81

 something

0.80

 mischief

0.80

 miscon

0.78

 hypocr

0.74

hay

0.73

 backwards

0.72

 misunderstood

0.72

Activations Density 0.471%

instances of actions or statements that could be seen as harmful or disruptive

phrases that convey action or inquiry

No Comments

No Known Activations