INDEX

Explanations

news articles and headlines

oai_token-act-pair · gpt-3.5-turbo

capitalized proper nouns or names

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

ãĤ¼ãĤ¦ãĤ¹

-0.79

ģĸ

-0.76

 differe

-0.72

 diplom

-0.70

é¾įå¥ĳå£«

-0.68

 GOODMAN

-0.68

æ©

-0.68

adm

-0.67

ãĤ´ãĥ³

-0.67

schild

-0.66

POSITIVE LOGITS

aired

1.25

ossession

1.24

redict

1.24

ossible

1.17

ardon

1.16

ierce

1.14

ulse

1.14

icking

1.13

odcast

1.12

uls

1.12

Activations Density 0.036%

news articles and headlines

capitalized proper nouns or names

No Comments

No Known Activations