INDEX

Explanations

sequences that signal the start or end of an advertisement

oai_token-act-pair · gpt-3.5-turbo

symbols or indicators of positive quantities or increments

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 1-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.1.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.1.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

 Grimes

-0.75

anium

-0.73

 Lauder

-0.65

 Expend

-0.64

ctuary

-0.63

 Sutherland

-0.62

essee

-0.62

 reintrodu

-0.62

NER

-0.61

ridor

-0.61

POSITIVE LOGITS

/-

1.24

-+

1.04

/+

0.97

--+

0.88

plus

0.80

events

0.79

cum

0.77

Pg

0.76

auto

0.73

IQ

0.72

Activations Density 0.010%

sequences that signal the start or end of an advertisement

symbols or indicators of positive quantities or increments

No Comments

No Known Activations