INDEX

Explanations

highly polarized and emotional statements

oai_token-act-pair · gpt-3.5-turbo

phrases with repeated symbols or unusual characters

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 9-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.9.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.9.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

 Panc

-0.68

entary

-0.66

 airs

-0.65

mares

-0.65

 pageant

-0.63

 Doodle

-0.62

iewicz

-0.62

 Alive

-0.62

 Ascension

-0.62

 Belg

-0.62

POSITIVE LOGITS

ª

1.31

ł

1.12

Ĳ

1.12

¹

1.11

«

1.10

ĳ

1.09

¤

1.09

¡

1.06

¦

1.03

®

1.00

Activations Density 0.145%

highly polarized and emotional statements

phrases with repeated symbols or unusual characters

No Comments

No Known Activations