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

 Crit

-0.70

assic

-0.66

Jag

-0.66

 Tracker

-0.65

Ard

-0.65

 Chart

-0.64

avascript

-0.64

ential

-0.63

 Notification

-0.62

 Mandatory

-0.61

POSITIVE LOGITS

awi

0.79

EStream

0.76

ãģ¦

0.73

ãĥ¼ãĥĨãĤ£

0.73

ĸļ

0.71

¬¼

0.71

ãģł

0.70

ette

0.70

ãģª

0.70

Activations Density 0.000%

No Known Activations

This feature has no known activations.