INDEX

Explanations

proper nouns related to news or media headlines

oai_token-act-pair · gpt-3.5-turbo

instances of the word "NEW" indicating a possible focus on new information or updates

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

stood

-0.82

ppe

-0.81

-0.68

cephal

-0.65

mop

-0.64

agn

-0.63

uca

-0.63

minecraft

-0.63

 McGee

-0.62

osate

-0.62

POSITIVE LOGITS

 YORK

1.39

foundland

1.13

PORT

1.10

ARK

0.96

CAST

0.93

 Orleans

0.89

ITY

0.88

bie

0.88

TON

0.86

MAN

0.84

Activations Density 0.008%

proper nouns related to news or media headlines

instances of the word "NEW" indicating a possible focus on new information or updates

No Comments

No Known Activations