INDEX

Explanations

phrases related to some kind of political reporting or commentary

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

 comprom

-0.68

oint

-0.61

ONSORED

-0.57

nib

-0.56

 toxin

-0.55

 respons

-0.54

fib

-0.52

ãĥ¥

-0.52

 direct

-0.52

 Tamil

-0.52

POSITIVE LOGITS

abouts

1.34

upon

1.18

FORE

0.97

fore

0.96

after

0.91

are

0.86

 Goes

0.86

ain

0.84

 Were

0.83

buquerque

0.83

Activations Density 0.092%

phrases related to some kind of political reporting or commentary

No Comments

No Known Activations