INDEX

Explanations

words ending with the suffix "-ion"

oai_token-act-pair · gpt-3.5-turbo

instances of the letter "l" 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 @ 10-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.10.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.10.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

 Shack

-0.68

 Subtle

-0.62

 Credit

-0.62

 centrif

-0.62

 backdrop

-0.61

 Bots

-0.61

 lawy

-0.59

 Vald

-0.59

 Belt

-0.58

wip

-0.58

POSITIVE LOGITS

ibrarian

1.47

adders

1.37

ibr

1.30

ivery

1.28

izards

1.25

ongh

1.24

asso

1.22

onel

1.21

ovable

1.18

aments

1.18

Activations Density 0.019%

words ending with the suffix "-ion"

instances of the letter "l" in various contexts

No Comments

No Known Activations