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

 balloon

-0.66

 rods

-0.65

rod

-0.64

 balloons

-0.63

 bubbles

-0.62

MEP

-0.62

upd

-0.61

pin

-0.61

DX

-0.60

rod

-0.60

POSITIVE LOGITS

illard

0.92

illance

0.81

ftime

0.77

oples

0.77

liber

0.72

wich

0.71

 occas

0.71

orest

0.70

arag

0.70

 conflic

0.69

Activations Density 0.000%

No Known Activations

This feature has no known activations.