INDEX

Explanations

numerical values embedded within words

oai_token-act-pair · gpt-3.5-turbo

occurrences of the substring "od"

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 1-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.1.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.1.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

Ago

-0.74

riott

-0.66

 Citadel

-0.64

 silence

-0.64

PAL

-0.63

 Quinn

-0.63

 Scand

-0.62

Ń·

-0.61

 Falcon

-0.61

 Crest

-0.60

POSITIVE LOGITS

od

1.34

odon

1.21

yssey

1.13

iamond

1.10

amn

1.08

sworth

1.06

wig

1.05

opter

1.04

unn

0.99

ata

0.99

Activations Density 0.009%

numerical values embedded within words

occurrences of the substring "od"

No Comments

No Known Activations