INDEX

Explanations

No Explanations Found

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 2-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.2.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.2.hook_resid_pre

Architecture

standard

Context Size

128

Dataset

Skylion007/openwebtext

Hook Point Layer

Activation Function

relu

Embeds

IFrame

Link

Not in Any Lists

No Comments

Negative Logits

 congratulated

-0.67

 encour

-0.65

 tracks

-0.64

 disbel

-0.64

ebook

-0.61

ynski

-0.61

hene

-0.60

 awaited

-0.59

fruit

-0.59

 trance

-0.59

POSITIVE LOGITS

 AUTH

0.73

andro

0.71

izza

0.71

iture

0.67

itect

0.67

 Parables

0.64

ature

0.63

Gameplay

0.63

 Icar

0.63

patrick

0.62

Activations Density 0.000%

No Known Activations

This feature has no known activations.