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

PlotsExplanationShow Test FieldDefault Test Text

IFrame

Link

Not in Any Lists

No Comments

Negative Logits

venture

-0.70

MAT

-0.65

CVE

-0.65

vg

-0.65

ALS

-0.64

cence

-0.64

arine

-0.64

HEAD

-0.63

pport

-0.62

fman

-0.62

POSITIVE LOGITS

minster

0.72

uder

0.67

 Offense

0.64

 Idol

0.61

attering

0.60

 Efficiency

0.60

ãĥĨ

0.60

Dyn

0.58

 Dian

0.58

Bei

0.58

Activations Density 0.000%

No Known Activations

This feature has no known activations.