INDEX

Explanations

references to different parts or sections of a text or document

oai_token-act-pair · gpt-3.5-turbo

references to different parts or sections of a document

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 8-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.8.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.8.hook_resid_pre

Architecture

standard

Context Size

128

Dataset

Skylion007/openwebtext

Hook Point Layer

Activation Function

relu

Embeds

IFrame

Link

Not in Any Lists

No Comments

Negative Logits

anca

-0.65

apons

-0.63

berus

-0.62

 speakers

-0.60

 gaping

-0.57

 minded

-0.56

asting

-0.56

ãĥīãĥ©ãĤ´ãĥ³

-0.56

âĤ¬

-0.56

acha

-0.55

POSITIVE LOGITS

ners

1.26

nered

1.22

icular

1.22

isans

1.16

icularly

1.16

ridge

1.15

icles

1.11

ially

1.10

icle

1.06

ner

1.05

Activations Density 0.038%

references to different parts or sections of a text or document

references to different parts or sections of a document

No Comments

No Known Activations