INDEX

Explanations

topics or keywords in a list

oai_token-act-pair · gpt-3.5-turbo

sections discussing various topics

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 1-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.1.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.1.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

aping

-0.74

 promise

-0.74

 ramp

-0.74

 abort

-0.73

 alliance

-0.70

 mete

-0.67

ades

-0.66

ter

-0.66

 testament

-0.65

vow

-0.65

POSITIVE LOGITS

Topics

1.95

afety

1.03

Topic

0.99

natureconservancy

0.94

ĸļ

0.93

matter

0.88

DragonMagazine

0.87

ICS

0.85

ãĤ¼ãĤ¦ãĤ¹

0.83

 Topics

0.81

Activations Density 0.013%

topics or keywords in a list

sections discussing various topics

No Comments

No Known Activations