INDEX

Explanations

phrases indicating contrast or exceptions

oai_token-act-pair · gpt-3.5-turbo

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

adian

-0.71

matic

-0.66

 scope

-0.64

shire

-0.57

raviolet

-0.56

pez

-0.56

izo

-0.56

 latitude

-0.55

FUL

-0.55

atu

-0.53

POSITIVE LOGITS

all

1.61

all

1.48

All

1.33

ALL

1.27

All

1.19

alls

1.00

ali

0.93

ALL

0.91

 whole

0.86

allo

0.78

Activations Density 0.368%

phrases indicating contrast or exceptions

No Comments

No Known Activations