INDEX

Explanations

instances where an action is recommended or instructed

oai_token-act-pair · gpt-3.5-turbo

statements that introduce or emphasize concepts or information

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 10-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.10.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.10.hook_resid_pre

Architecture

standard

Context Size

128

Dataset

Skylion007/openwebtext

Hook Point Layer

Activation Function

relu

Embeds

IFrame

Link

Not in Any Lists

No Comments

Negative Logits

Ĭ±

-0.76

aths

-0.69

 Streets

-0.68

chev

-0.68

farious

-0.67

ARS

-0.66

aneers

-0.66

»Ĵ

-0.66

rils

-0.66

amia

-0.64

POSITIVE LOGITS

 ensures

1.41

 means

1.39

 allows

1.33

 enables

1.29

 implies

1.28

 translates

1.28

 eliminates

1.24

 includes

1.24

 prevents

1.24

 entails

1.21

Activations Density 0.146%

instances where an action is recommended or instructed

statements that introduce or emphasize concepts or information

No Comments

No Known Activations