INDEX

Explanations

names or titles with the initials "E" followed by a single-digit number

oai_token-act-pair · gpt-3.5-turbo

references to individuals with the initial "E."

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

 alleg

-0.65

 chaos

-0.64

 louder

-0.63

 pandemonium

-0.62

 quotations

-0.62

Aph

-0.62

 Equality

-0.61

 possibility

-0.61

 props

-0.60

 Korra

-0.59

POSITIVE LOGITS

isner

1.38

ileen

1.32

riks

1.31

rika

1.23

pps

1.22

ller

1.22

nerg

1.18

verson

1.18

asley

1.18

thel

1.16

Activations Density 0.022%

names or titles with the initials "E" followed by a single-digit number

references to individuals with the initial "E."

No Comments

No Known Activations