INDEX

Explanations

phrases related to action or activity

oai_token-act-pair · gpt-3.5-turbo

phrases indicating problematic situations or actions

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

IFrame

Link

Not in Any Lists

No Comments

Negative Logits

onga

-0.60

onto

-0.59

aka

-0.54

iaz

-0.53

 wearer

-0.48

orio

-0.48

interstitial

-0.47

 hereafter

-0.47

currently

-0.46

cko

-0.46

POSITIVE LOGITS

 yesterday

0.64

 earlier

0.55

 last

0.55

 awaken

0.52

 initially

0.51

 originally

0.50

ãĢĲ

0.50

cffffcc

0.50

terday

0.50

 Yanuk

0.50

Activations Density 3.504%

phrases related to action or activity

phrases indicating problematic situations or actions

No Comments

No Known Activations