INDEX

Explanations

No Explanations Found

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

IFrame

Link

Not in Any Lists

No Comments

Negative Logits

 lasting

-0.80

 denomin

-0.73

 Shap

-0.66

 univers

-0.64

 centered

-0.64

 ranged

-0.63

 subst

-0.62

 drawn

-0.62

 favor

-0.61

 universal

-0.61

POSITIVE LOGITS

£ı

0.85

milo

0.85

itored

0.83

PsyNetMessage

0.81

Offline

0.78

Filename

0.77

eering

0.76

wcsstore

0.76

soDeliveryDate

0.76

romeda

0.75

Activations Density 0.000%

No Known Activations

This feature has no known activations.