INDEX

Explanations

scientific or technical terms characterized by specific descriptions or definitions

oai_token-act-pair · gpt-3.5-turbo

phrases that describe characteristics or defining traits of various subjects

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

adra

-0.85

aea

-0.80

nuts

-0.73

heed

-0.72

worth

-0.70

hiba

-0.70

loo

-0.68

ammy

-0.67

BA

-0.67

isdom

-0.67

POSITIVE LOGITS

 characterized

0.80

urally

0.72

ocument

0.72

REDACTED

0.71

%]

0.71

ATING

0.69

ounces

0.68

xual

0.68

idable

0.67

istics

0.67

Activations Density 0.019%

scientific or technical terms characterized by specific descriptions or definitions

phrases that describe characteristics or defining traits of various subjects

No Comments

No Known Activations