INDEX

Explanations

words with 'h' followed by a number, potentially representing errors or codes in a technical context

oai_token-act-pair · gpt-3.5-turbo

occurrences of the letter 'h'

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 9-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.9.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.9.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

 Nightmares

-0.64

 merc

-0.61

 presses

-0.60

 Tyrann

-0.59

 Appropriations

-0.58

 Suns

-0.58

iatrics

-0.57

 pandemonium

-0.57

glomer

-0.57

 Revival

-0.56

POSITIVE LOGITS

adow

1.22

ield

1.15

irts

1.10

ooter

1.07

IFT

1.00

oulder

0.99

uffle

0.99

ifting

0.99

ooters

0.99

aders

0.98

Activations Density 0.044%

words with 'h' followed by a number, potentially representing errors or codes in a technical context

occurrences of the letter 'h'

No Comments

No Known Activations