INDEX

Explanations

references to specific historical eras

oai_token-act-pair · gpt-3.5-turbo

references to different historical or cultural eras

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 4-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.4.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.4.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

lations

-0.80

acts

-0.78

 drawer

-0.74

gments

-0.74

atem

-0.74

amina

-0.74

ramid

-0.73

portion

-0.72

tein

-0.72

Ĥ¬

-0.71

POSITIVE LOGITS

era

0.98

Era

0.89

 epoch

0.88

icals

0.86

UTC

0.83

 nostalg

0.76

hetti

0.71

osaurs

0.69

Age

0.69

 Invention

0.68

Activations Density 0.017%

references to specific historical eras

references to different historical or cultural eras

No Comments

No Known Activations