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

sid

-0.72

cffff

-0.66

 Sachs

-0.65

phe

-0.63

 commissioner

-0.63

redited

-0.61

Ph

-0.61

raid

-0.61

asper

-0.59

 Commissioner

-0.59

POSITIVE LOGITS

blance

0.70

ellation

0.67

 enrichment

0.67

wear

0.67

uania

0.66

Design

0.66

enges

0.65

estate

0.65

wcsstore

0.65

Appearance

0.63

Activations Density 0.000%

No Known Activations

This feature has no known activations.