INDEX

Explanations

proper nouns or other specific words related to certain entities or events

oai_token-act-pair · gpt-3.5-turbo

line breaks or breaks in text formatting

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 8-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.8.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.8.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

WARE

-0.72

balls

-0.67

meal

-0.66

eers

-0.64

Sax

-0.63

 catch

-0.62

tale

-0.61

 Totem

-0.59

 caught

-0.58

ãģŁ

-0.57

POSITIVE LOGITS

ackets

1.28

acket

1.14

anches

1.06

anch

1.04

igham

1.04

unn

1.01

aces

0.99

OAD

0.99

ushed

0.97

acing

0.95

Activations Density 0.017%

proper nouns or other specific words related to certain entities or events

line breaks or breaks in text formatting

No Comments

No Known Activations