INDEX

Explanations

words related to content going viral on the internet

oai_token-act-pair · gpt-3.5-turbo

references to content that has gone viral

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

sterdam

-0.79

yer

-0.78

quartered

-0.76

eln

-0.75

zzo

-0.75

fman

-0.72

ebus

-0.72

ricting

-0.70

osher

-0.70

Ã±a

-0.68

POSITIVE LOGITS

 Haram

0.83

 viral

0.81

aneously

0.80

 sensation

0.80

iously

0.79

iform

0.71

film

0.71

upd

0.70

 ILCS

0.69

ously

0.68

Activations Density 0.016%

words related to content going viral on the internet

references to content that has gone viral

No Comments

No Known Activations