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

PlotsExplanationShow Test FieldDefault Test Text

IFrame

Link

Not in Any Lists

No Comments

Negative Logits

emonic

-0.78

oric

-0.76

amination

-0.74

ures

-0.74

anes

-0.71

amines

-0.71

phabet

-0.70

ptic

-0.69

omic

-0.68

Abstract

-0.68

POSITIVE LOGITS

 ãĤµãĥ¼ãĥĨãĤ£ãĥ¯ãĥ³

0.86

ãĥ´

0.84

ãĥ¼ãĥĨãĤ£

0.82

 WARN

0.78

Interstitial

0.75

lvl

0.75

çĶ°

0.74

iHUD

0.73

æ©

0.73

ãĥĨãĤ£

0.70

Activations Density 0.000%

No Known Activations

This feature has no known activations.