INDEX

Explanations

No Explanations Found

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 3-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.3.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.3.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

overe

-0.71

oreAnd

-0.65

 Depot

-0.64

 Gleaming

-0.63

 Luxem

-0.62

ĨĴ

-0.62

 confir

-0.61

 achievement

-0.61

SIG

-0.61

 shattered

-0.61

POSITIVE LOGITS

TPPStreamerBot

0.74

lan

0.69

ichick

0.65

SPONSORED

0.64

ially

0.63

bender

0.63

Fi

0.63

GIF

0.62

Cs

0.61

 Rapp

0.60

Activations Density 0.000%

No Known Activations

This feature has no known activations.