INDEX

Explanations

proper nouns possibly related to entertainment or tech

oai_token-act-pair · gpt-3.5-turbo

names of well-known personalities or brands related to media and entertainment

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 2-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.2.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.2.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

etooth

-0.98

aird

-0.87

olulu

-0.86

cham

-0.86

quarters

-0.84

neapolis

-0.83

vertis

-0.83

hari

-0.81

hammad

-0.81

zac

-0.80

POSITIVE LOGITS

 Claus

0.76

 Winc

0.72

 Canary

0.71

 Constantine

0.69

 volts

0.68

 Flint

0.68

 Faust

0.67

 Valiant

0.66

 Candle

0.66

 MÃ©

0.66

Activations Density 0.016%

proper nouns possibly related to entertainment or tech

names of well-known personalities or brands related to media and entertainment

No Comments

No Known Activations