INDEX

Explanations

phrases related to strong opinions or stances on various topics

oai_token-act-pair · gpt-3.5-turbo

phrases or terms that support a particular viewpoint or ideology

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

IFrame

Link

Not in Any Lists

No Comments

Negative Logits

ĸļ

-0.91

ãĤ¼ãĤ¦ãĤ¹

-0.77

 Halls

-0.68

 Sins

-0.65

 Gorge

-0.65

inia

-0.64

 Dickens

-0.62

 Twain

-0.62

 Brooks

-0.60

 pains

-0.60

POSITIVE LOGITS

digy

1.45

dding

1.27

actively

1.26

verbs

1.16

pelling

1.15

dig

1.11

ccess

1.10

strate

1.09

gressive

1.06

ctor

1.06

Activations Density 0.015%

phrases related to strong opinions or stances on various topics

phrases or terms that support a particular viewpoint or ideology

No Comments

No Known Activations