INDEX

Explanations

No Explanations Found

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 0-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.0.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.0.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

isu

-0.76

cookie

-0.73

jab

-0.72

gs

-0.72

visor

-0.72

boxing

-0.72

culosis

-0.70

lake

-0.70

abyte

-0.67

frog

-0.67

POSITIVE LOGITS

 Apprentice

0.79

 Commun

0.70

Assembly

0.65

TOD

0.65

NEC

0.64

 athe

0.63

 Ernst

0.63

 millenn

0.63

 Brun

0.62

iliated

0.62

Activations Density 0.000%

No Known Activations

This feature has no known activations.