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

 Kenobi

-0.68

 Ascend

-0.67

 Prometheus

-0.65

actor

-0.65

 Babylon

-0.65

 Bots

-0.64

jong

-0.64

stellar

-0.63

 Redemption

-0.63

 perenn

-0.61

POSITIVE LOGITS

zinski

0.94

å¥³

0.78

ãĤ¢ãĥ«

0.73

ANS

0.72

itol

0.69

anca

0.68

arte

0.68

Posted

0.67

ÑĮ

0.67

icone

0.66

Activations Density 0.000%

No Known Activations

This feature has no known activations.