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

IFrame

Link

Not in Any Lists

No Comments

Negative Logits

 sovere

-0.91

othe

-0.66

ographical

-0.65

 unexpl

-0.64

earchers

-0.62

 attractions

-0.62

 constitu

-0.60

olog

-0.59

asonry

-0.59

 embarrassment

-0.58

POSITIVE LOGITS

paren

0.66

ãĤ¼ãĤ¦ãĤ¹

0.64

calling

0.64

nces

0.62

ARP

0.62

ions

0.61

 tripled

0.61

kj

0.61

alion

0.61

Balance

0.61

Activations Density 0.000%

No Known Activations

This feature has no known activations.