INDEX

Explanations

No Explanations Found

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 3-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.3.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.3.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

terday

-0.94

 Amen

-0.70

omas

-0.68

marks

-0.68

lights

-0.67

 Rats

-0.65

rows

-0.63

theless

-0.63

'/

-0.63

lier

-0.62

POSITIVE LOGITS

srf

0.71

Obj

0.69

chio

0.69

anooga

0.67

Crash

0.66

©¶æ¥µ

0.66

iPhone

0.65

onz

0.63

 adolesc

0.63

 lapt

0.62

Activations Density 0.000%

No Known Activations

This feature has no known activations.