INDEX

Explanations

phrases related to consequences or methods

oai_token-act-pair · gpt-3.5-turbo

phrases indicating significance or implications of various subjects

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

reb

-0.72

edit

-0.65

greg

-0.59

alted

-0.59

older

-0.59

rex

-0.58

 Bene

-0.57

rage

-0.56

ersen

-0.56

icent

-0.56

POSITIVE LOGITS

 means

3.65

 Means

2.62

 meant

1.92

 mean

1.81

 entails

1.53

 signifies

1.52

 implies

1.50

 translates

1.49

 equals

1.32

 denotes

1.24

Activations Density 0.027%

phrases related to consequences or methods

phrases indicating significance or implications of various subjects

No Comments

No Known Activations