INDEX

Explanations

phrases related to being open, unrestricted, or limitless

oai_token-act-pair · gpt-3.5-turbo

references to concepts of openness and transparency in various contexts

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 11-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.11.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.11.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

 Brach

-0.72

 Mons

-0.70

 Tsarnaev

-0.69

 Lumpur

-0.69

Mub

-0.68

 Plum

-0.68

Tik

-0.66

 Comet

-0.66

 Bucc

-0.66

UD

-0.65

POSITIVE LOGITS

ended

1.45

source

1.40

minded

1.23

access

1.13

concept

1.12

air

1.11

world

1.11

bodied

1.09

eyed

1.09

door

1.08

Activations Density 0.023%

phrases related to being open, unrestricted, or limitless

references to concepts of openness and transparency in various contexts

No Comments

No Known Activations