INDEX

Explanations

notable or highlighted information or entities

oai_token-act-pair · gpt-3.5-turbo

phrases that emphasize notable events or features

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 1-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.1.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.1.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

tein

-0.89

rieving

-0.78

elf

-0.77

sett

-0.72

ual

-0.72

izoph

-0.70

rive

-0.70

rieve

-0.70

Haunted

-0.70

CDC

-0.70

POSITIVE LOGITS

 notably

0.82

 absent

0.78

 notable

0.75

 noteworthy

0.75

 Vict

0.70

 shines

0.69

fav

0.68

 noticeable

0.68

 conduc

0.67

 prominently

0.67

Activations Density 0.005%

notable or highlighted information or entities

phrases that emphasize notable events or features

No Comments

No Known Activations