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

tor

-0.79

 Table

-0.67

axe

-0.67

ison

-0.66

ardless

-0.66

ciating

-0.66

ormon

-0.63

paralle

-0.62

izont

-0.62

igslist

-0.61

POSITIVE LOGITS

 GOODMAN

0.93

jad

0.67

os

0.65

Cth

0.65

MpServer

0.65

 juveniles

0.64

 susp

0.64

 salute

0.63

 kidding

0.63

 distressed

0.62

Activations Density 0.000%

No Known Activations

This feature has no known activations.