INDEX

Explanations

phrases indicating contrast or contradiction

oai_token-act-pair · gpt-3.5-turbo

negation and phrases that express conditional relationships

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 5-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.5.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.5.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

chnology

-0.57

å£«

-0.55

²¾

-0.55

auri

-0.54

available

-0.54

iHUD

-0.52

cms

-0.52

ricks

-0.51

cream

-0.51

gil

-0.51

POSITIVE LOGITS

 Xander

0.53

THR

0.51

 apologies

0.49

ãĤĵ

0.46

 Woodward

0.45

 Miliband

0.45

 Opinion

0.44

 Ender

0.44

 congr

0.43

 pardon

0.43

Activations Density 1.114%

phrases indicating contrast or contradiction

negation and phrases that express conditional relationships

No Comments

No Known Activations