INDEX

Explanations

the letter "E" followed by a number

oai_token-act-pair · gpt-3.5-turbo

mentions of the Eiffel Tower

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 5-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.5.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.5.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

 Kenobi

-0.76

 Anthem

-0.70

 Showdown

-0.65

ãĤ¤ãĥĪ

-0.65

wagen

-0.64

 pains

-0.63

 Akira

-0.62

 Nickel

-0.61

 Bleach

-0.61

raints

-0.60

POSITIVE LOGITS

nerg

1.07

fficient

1.06

ighty

1.04

isner

1.04

rect

1.03

ASY

1.00

lev

1.00

AST

1.00

resp

1.00

uter

1.00

Activations Density 0.030%

the letter "E" followed by a number

mentions of the Eiffel Tower

No Comments

No Known Activations