INDEX

Explanations

references to a specific location, possibly related to a place called "Waukegan."

oai_token-act-pair · gpt-3.5-turbo

the repeated mention of the term "au" in various contexts

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

appers

-0.68

eyes

-0.64

OB

-0.63

 Cavaliers

-0.60

Io

-0.60

APS

-0.60

orative

-0.60

 Attribution

-0.59

STATE

-0.59

bread

-0.59

POSITIVE LOGITS

llah

1.25

gment

1.13

ction

1.07

pload

1.00

fman

1.00

qua

0.99

cham

0.98

cel

0.95

ghan

0.93

lly

0.92

Activations Density 0.027%

references to a specific location, possibly related to a place called "Waukegan."

the repeated mention of the term "au" in various contexts

No Comments

No Known Activations