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

Tele

-0.72

 secretaries

-0.67

izza

-0.67

irgin

-0.66

dq

-0.66

 clerics

-0.62

rotein

-0.61

afer

-0.61

 Dull

-0.60

Delivery

-0.60

POSITIVE LOGITS

taboola

0.75

 Skywalker

0.68

âķĲ

0.68

 ILCS

0.67

TMZ

0.65

Tat

0.64

 decomp

0.64

ocene

0.64

olson

0.62

 Starcraft

0.62

Activations Density 0.000%

No Known Activations

This feature has no known activations.