INDEX

Explanations

references to additional information or related topics

oai_token-act-pair · gpt-3.5-turbo

references to additional content or sources

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

angan

-0.90

ufact

-0.70

heit

-0.66

vernight

-0.65

atform

-0.63

 Shutdown

-0.63

issance

-0.62

 depended

-0.60

tein

-0.60

mort

-0.59

POSITIVE LOGITS

paren

0.69

unal

0.68

onnaissance

0.67

 similarities

0.67

 footprints

0.66

 afar

0.65

Spread

0.65

ideos

0.65

xual

0.65

 limits

0.65

Activations Density 0.126%

references to additional information or related topics

references to additional content or sources

No Comments

No Known Activations