INDEX

Explanations

No Explanations Found

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 1-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.1.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.1.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

 Categories

-0.69

 Commands

-0.69

 Moto

-0.68

XV

-0.65

Ans

-0.62

 Enlight

-0.62

ictional

-0.62

 Variable

-0.61

igue

-0.60

 Aber

-0.59

POSITIVE LOGITS

rent

0.75

duction

0.72

Bloomberg

0.71

mining

0.69

icter

0.68

mission

0.67

atomic

0.67

ween

0.66

uggets

0.66

mitt

0.66

Activations Density 0.000%

No Known Activations

This feature has no known activations.