INDEX

Explanations

mentions of the word "mag" at various context

oai_token-act-pair · gpt-3.5-turbo

references to magma

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 0-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.0.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.0.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

_>

-0.78

 Zuckerberg

-0.75

ourcing

-0.66

 Bernstein

-0.65

 Crossing

-0.65

ospons

-0.64

bda

-0.64

 BlackBerry

-0.63

 Bean

-0.62

Assad

-0.62

POSITIVE LOGITS

mag

4.01

mag

2.51

Mag

2.45

Mag

2.40

MAG

1.73

MAG

1.53

 magazines

1.45

 magician

1.43

 mage

1.34

 mages

1.33

Activations Density 0.020%

mentions of the word "mag" at various context

references to magma

No Comments

No Known Activations