INDEX

Explanations

keywords related to having permission or ability to use something

oai_token-act-pair · gpt-3.5-turbo

instances of the word "access" 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

 Kern

-0.59

 grandmother

-0.56

kos

-0.55

-0.54

Jou

-0.53

 cousin

-0.53

 candle

-0.53

Lens

-0.52

 circling

-0.52

 colleague

-0.52

POSITIVE LOGITS

ibly

1.28

ibility

1.28

 thereto

1.18

to

1.11

ible

1.04

ioned

0.97

ibilities

0.94

ibl

0.88

ion

0.86

IBLE

0.83

Activations Density 0.030%

keywords related to having permission or ability to use something

instances of the word "access" in various contexts

No Comments

No Known Activations