INDEX

Explanations

No Explanations Found

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 1-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.1.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.1.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

iles

-0.85

ortium

-0.69

oulos

-0.68

iler

-0.66

owan

-0.64

SHE

-0.64

 distributions

-0.63

 contrace

-0.62

ierre

-0.62

 reconc

-0.61

POSITIVE LOGITS

deck

0.71

kept

0.68

POST

0.68

ception

0.66

ettle

0.66

dig

0.65

 Myster

0.64

pee

0.62

APTER

0.62

 Psychiat

0.62

Activations Density 0.000%

No Known Activations

This feature has no known activations.