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

 lapt

-0.80

 charact

-0.73

 toile

-0.71

 behav

-0.71

cffffcc

-0.70

 surv

-0.70

utory

-0.69

uten

-0.68

 oath

-0.67

_-

-0.67

POSITIVE LOGITS

 Carm

0.73

¶æ

0.73

Hos

0.67

æŃ

0.67

istries

0.64

 Galile

0.64

éĥ

0.62

å

0.62

 Robo

0.60

 Canaver

0.60

Activations Density 0.000%

No Known Activations

This feature has no known activations.