INDEX

Explanations

references to stickers or labels

oai_token-act-pair · gpt-3.5-turbo

references to stickers and their various uses or descriptions

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 0-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.0.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.0.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

akespeare

-0.75

RAW

-0.69

Rus

-0.68

 Pagan

-0.66

 succession

-0.66

 River

-0.66

rawdownloadcloneembedreportprint

-0.66

River

-0.65

apeake

-0.65

urnal

-0.65

POSITIVE LOGITS

 sticker

3.90

 stickers

3.16

 bumper

1.62

 adhesive

1.23

 stamp

1.22

 badge

1.21

stick

1.19

 plaque

1.16

 billboard

1.14

 stamps

1.13

Activations Density 0.033%

references to stickers or labels

references to stickers and their various uses or descriptions

No Comments

No Known Activations