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

IFrame

Link

Not in Any Lists

No Comments

Negative Logits

ãĤ©

-0.86

 Koen

-0.69

lihood

-0.68

ãĤ¬

-0.61

½

-0.60

 hepat

-0.60

 Wein

-0.59

nown

-0.59

 ãĤµãĥ¼ãĥĨãĤ£ãĥ¯ãĥ³

-0.58

 Bounty

-0.57

POSITIVE LOGITS

 curated

0.64

 comma

0.64

oday

0.63

maid

0.62

scr

0.61

enced

0.61

urer

0.61

ographies

0.61

 UNCLASSIFIED

0.60

taboola

0.59

Activations Density 0.000%

No Known Activations

This feature has no known activations.