INDEX

Explanations

calls to action or instructions to interact with digital content

oai_token-act-pair · gpt-3.5-turbo

instructions for user interaction, such as closing or printing

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

ãĤ©

-0.71

orian

-0.65

èª

-0.62

ãĥķãĤ©

-0.62

ãĤ¼ãĤ¦ãĤ¹

-0.62

 lantern

-0.60

ãĥīãĥ©ãĤ´ãĥ³

-0.59

ULAR

-0.59

mar

-0.59

dos

-0.59

POSITIVE LOGITS

 Wish

0.61

ername

0.59

 Cancel

0.58

 Bulk

0.57

requisite

0.57

edIn

0.56

 Oops

0.56

Tee

0.55

 Username

0.54

 Password

0.53

Activations Density 0.049%

calls to action or instructions to interact with digital content

instructions for user interaction, such as closing or printing

No Comments

No Known Activations