INDEX

Explanations

phrases related to legal actions and statements

oai_token-act-pair · gpt-3.5-turbo

commanding phrases or direct instructions

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

pmwiki

-0.55

 incent

-0.49

abre

-0.47

 stocking

-0.47

 Azerb

-0.46

 accommodating

-0.45

£ı

-0.45

 millenn

-0.45

 comr

-0.45

 condu

-0.45

POSITIVE LOGITS

 ONLY

0.62

 doesn

0.59

it

0.57

 only

0.56

don

0.54

 hasn

0.54

 nobody

0.53

 hers

0.53

isn

0.52

only

0.52

Activations Density 0.975%

phrases related to legal actions and statements

commanding phrases or direct instructions

No Comments

No Known Activations