INDEX

Explanations

No Explanations Found

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 4-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.4.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.4.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

minster

-0.77

 enlarg

-0.65

-0.63

merce

-0.62

enezuel

-0.62

 reciproc

-0.62

ofi

-0.61

iod

-0.61

ieve

-0.61

Ñı

-0.60

POSITIVE LOGITS

 Pieces

0.62

Wheel

0.62

zed

0.61

Bal

0.61

PB

0.60

tes

0.60

sts

0.60

 Preferred

0.59

ttes

0.59

PB

0.58

Activations Density 0.000%

No Known Activations

This feature has no known activations.