INDEX

Explanations

references to very specific entities, in this case, focusing on variations involving "George W. Bush" and the number values are likely emphasizing different contexts or variations

oai_token-act-pair · gpt-3.5-turbo

mentions of George W. Bush

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 7-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.7.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.7.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

 heels

-0.64

liga

-0.62

mobi

-0.61

ocre

-0.60

xc

-0.59

wcs

-0.59

 complementary

-0.58

NRS

-0.58

organic

-0.58

 complimentary

-0.58

POSITIVE LOGITS

 Abrams

0.72

ynes

0.67

 Branch

0.64

 Lines

0.64

 Reck

0.63

 Bush

0.62

itton

0.62

 Romero

0.62

 Bates

0.58

agan

0.57

Activations Density 0.061%

references to very specific entities, in this case, focusing on variations involving "George W. Bush" and the number values are likely emphasizing different contexts or variations

mentions of George W. Bush

No Comments

No Known Activations