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

isans

-0.72

choice

-0.65

 judgment

-0.64

atform

-0.64

birth

-0.64

iform

-0.63

ascus

-0.63

²¾

-0.63

devices

-0.62

etsk

-0.62

POSITIVE LOGITS

Redd

0.72

 Berm

0.70

OSED

0.66

OND

0.65

 Yuan

0.65

racuse

0.65

ordes

0.64

 Chung

0.64

vance

0.60

ãĥĩãĤ£

0.60

Activations Density 0.000%

No Known Activations

This feature has no known activations.