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

 Cancel

-0.73

 proble

-0.65

 Awareness

-0.64

Alm

-0.64

 Uriel

-0.64

 Unsure

-0.62

ivia

-0.59

 upkeep

-0.59

Bei

-0.59

 disadvant

-0.58

POSITIVE LOGITS

for

0.85

 Bridgewater

0.68

riot

0.65

ylum

0.64

for

0.62

enton

0.61

atri

0.60

apon

0.60

uit

0.60

uka

0.60

Activations Density 0.000%

No Known Activations

This feature has no known activations.