INDEX

Explanations

phrases related to a person named "Pe" in various contexts

oai_token-act-pair · gpt-3.5-turbo

mentions of the term "Pe"

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 8-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.8.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.8.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

schild

-0.77

naire

-0.77

 EDITION

-0.76

stretched

-0.76

selves

-0.72

etheless

-0.69

doms

-0.65

Â¯

-0.65

 Balanced

-0.64

orage

-0.63

POSITIVE LOGITS

anut

1.40

anuts

1.28

asant

1.18

eking

1.08

pperc

1.06

ek

1.05

gging

1.04

eper

1.04

ggie

1.03

eps

1.02

Activations Density 0.015%

phrases related to a person named "Pe" in various contexts

mentions of the term "Pe"

No Comments

No Known Activations