INDEX

Explanations

phrases indicating a wide variety or inclusivity in statements

oai_token-act-pair · gpt-3.5-turbo

the phrase "pretty much" and variations thereof, indicating a generalizing or summarizing tone

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

IFrame

Link

Not in Any Lists

No Comments

Negative Logits

alin

-0.66

tnc

-0.62

aley

-0.62

 Hague

-0.61

odium

-0.61

 Revenue

-0.60

æĿ

-0.59

poke

-0.59

HI

-0.57

swick

-0.56

POSITIVE LOGITS

 everywhere

1.20

 every

1.13

 everything

1.13

 everyone

1.09

 everybody

1.08

 anything

1.07

 EVERY

1.07

everything

1.03

anything

1.02

every

1.01

Activations Density 0.082%

phrases indicating a wide variety or inclusivity in statements

the phrase "pretty much" and variations thereof, indicating a generalizing or summarizing tone

No Comments

No Known Activations