INDEX

Explanations

names of specific individuals

oai_token-act-pair · gpt-3.5-turbo

names and terms associated with individuals and families

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

hovah

-0.93

chall

-0.85

psey

-0.83

aneers

-0.81

neys

-0.80

edes

-0.78

role

-0.77

lyak

-0.77

ply

-0.77

ed

-0.77

POSITIVE LOGITS

Berry

0.82

cence

0.68

ERY

0.68

ULTS

0.66

 Gaga

0.64

Rey

0.64

ãĥ´ãĤ¡

0.64

 Glob

0.64

 Flavoring

0.63

artz

0.62

Activations Density 0.072%

names of specific individuals

names and terms associated with individuals and families

No Comments

No Known Activations