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

SHIP

-0.77

Ĺ

-0.71

 Gund

-0.68

Į

-0.67

gans

-0.66

 Fundamental

-0.64

 Gaia

-0.63

 Quit

-0.62

 Mongolia

-0.62

 counting

-0.61

POSITIVE LOGITS

itri

0.86

ether

0.85

hib

0.80

amara

0.78

issan

0.77

ilitary

0.77

emi

0.76

azine

0.75

 guiActiveUn

0.73

ovie

0.73

Activations Density 0.000%

No Known Activations

This feature has no known activations.