INDEX

Explanations

names or terms related to individuals or locations with 'El' as a prefix

oai_token-act-pair · gpt-3.5-turbo

references to specific individuals or names in the text

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

 IMAGES

-0.71

 dearly

-0.68

bom

-0.64

 indo

-0.62

okemon

-0.62

 senseless

-0.61

 captcha

-0.60

 exponent

-0.60

ACS

-0.60

 stunt

-0.59

POSITIVE LOGITS

abeth

0.91

otte

0.85

 Musk

0.75

quez

0.74

rette

0.73

abal

0.72

Raid

0.72

abet

0.71

Ree

0.71

onde

0.71

Activations Density 0.081%

names or terms related to individuals or locations with 'El' as a prefix

references to specific individuals or names in the text

No Comments

No Known Activations