INDEX

Explanations

references to specific parts or sections within something

oai_token-act-pair · gpt-3.5-turbo

references to specific components or parts of objects or concepts

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 10-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.10.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.10.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

 Skies

-0.65

 Rampage

-0.65

berus

-0.62

Fight

-0.61

 commissions

-0.59

ãĤī

-0.57

gently

-0.55

stasy

-0.54

 Surge

-0.54

balance

-0.54

POSITIVE LOGITS

ridge

1.26

ridges

1.23

icular

1.22

ials

1.12

aking

1.01

isans

1.01

icle

0.95

ners

0.94

icularly

0.93

ially

0.93

Activations Density 0.057%

references to specific parts or sections within something

references to specific components or parts of objects or concepts

No Comments

No Known Activations