INDEX

Explanations

questions or prompts directed at someone and often involving decision-making

oai_token-act-pair · gpt-3.5-turbo

the use of the word "you" in various contexts

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 4-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.4.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.4.hook_resid_pre

Architecture

standard

Context Size

128

Dataset

Skylion007/openwebtext

Hook Point Layer

Activation Function

relu

Embeds

PlotsExplanationShow Test FieldDefault Test Text

IFrame

Link

Not in Any Lists

No Comments

Negative Logits

ruciating

-0.77

icent

-0.71

ges

-0.69

Pg

-0.66

 Canaver

-0.65

ipal

-0.64

 Roose

-0.62

photos

-0.62

Kat

-0.62

peed

-0.62

POSITIVE LOGITS

 guys

1.24

're

1.12

tub

1.03

 want

0.94

've

0.94

 know

0.93

 wanna

0.91

 think

0.90

 intend

0.90

sir

0.87

Activations Density 0.080%

questions or prompts directed at someone and often involving decision-making

the use of the word "you" in various contexts

No Comments

No Known Activations