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

ãĢı

-0.83

bsite

-0.74

 metast

-0.72

GBT

-0.70

qqa

-0.66

querque

-0.66

 skelet

-0.66

 notor

-0.65

aukee

-0.65

Scotland

-0.63

POSITIVE LOGITS

 Enforcement

0.71

Yel

0.70

upper

0.69

ortunate

0.68

onomic

0.67

hement

0.66

ounce

0.66

ICAN

0.65

 Rican

0.63

arious

0.62

Activations Density 0.000%

No Known Activations

This feature has no known activations.