INDEX

Explanations

items or concepts that can be of varied types or purposes

oai_token-act-pair · gpt-3.5-turbo

terms indicating inclusivity or the ability to use various options

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 5-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.5.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.5.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

alus

-0.83

illard

-0.83

sson

-0.71

SA

-0.69

mand

-0.67

irth

-0.65

BA

-0.65

opsis

-0.63

pu

-0.63

pee

-0.62

POSITIVE LOGITS

 imaginable

1.22

THING

1.08

 conceivable

1.00

where

0.97

body

0.95

 else

0.89

 sorts

0.87

 kind

0.85

thin

0.82

 sort

0.78

Activations Density 0.064%

items or concepts that can be of varied types or purposes

terms indicating inclusivity or the ability to use various options

No Comments

No Known Activations