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

spo

-0.67

 Spit

-0.64

âĶĢ

-0.63

 Blitz

-0.63

 torn

-0.61

eers

-0.61

Ship

-0.61

DCS

-0.61

+++

-0.60

 Leviathan

-0.59

POSITIVE LOGITS

osexual

0.79

merce

0.77

isexual

0.74

opez

0.71

mobi

0.70

isoft

0.68

wal

0.66

],[

0.66

ynski

0.66

apps

0.64

Activations Density 0.000%

No Known Activations

This feature has no known activations.