INDEX

Explanations

phrases related to statistics or numerical quantities

oai_token-act-pair · gpt-3.5-turbo

phrases indicating frequency or degree of occurrence

oai_token-act-pair · gpt-4o-mini Triggered by @bot

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 3-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.3.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.3.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

Ö

-0.81

ocratic

-0.78

umbn

-0.67

×¨

-0.65

åĤ

-0.64

oder

-0.64

ensis

-0.64

alian

-0.64

plings

-0.64

è¦ļéĨĴ

-0.64

POSITIVE LOGITS

 Stories

0.73

estine

0.71

 Enough

0.70

entimes

0.70

 Helpful

0.69

ths

0.68

 Leader

0.68

 Says

0.67

 Than

0.67

Sounds

0.67

Activations Density 0.019%

phrases related to statistics or numerical quantities

phrases indicating frequency or degree of occurrence

No Comments

No Known Activations