INDEX

Explanations

names of specific individuals

oai_token-act-pair · gpt-3.5-turbo

proper nouns, particularly names and titles

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 11-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.11.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.11.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

IUM

-0.85

SHARE

-0.81

 SHARES

-0.76

INST

-0.71

 OFFIC

-0.68

FUN

-0.68

 NEXT

-0.68

SIZE

-0.66

 FACE

-0.63

FACE

-0.63

POSITIVE LOGITS

illard

1.06

arre

0.96

atson

0.90

quart

0.88

ennett

0.86

oldemort

0.85

opez

0.84

axter

0.83

iggs

0.82

arrett

0.81

Activations Density 0.089%

names of specific individuals

proper nouns, particularly names and titles

No Comments

No Known Activations