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

rust

-0.73

ato

-0.72

idd

-0.64

atari

-0.62

ighed

-0.62

rain

-0.61

uki

-0.60

ÃŃa

-0.59

cu

-0.59

uit

-0.58

POSITIVE LOGITS

²¾

0.80

 guiActiveUn

0.76

dayName

0.75

CLASSIFIED

0.73

ĻĤ

0.71

WARE

0.71

cgi

0.70

ĵĺ

0.67

 handc

0.66

©¶æ¥µ

0.66

Activations Density 0.000%

No Known Activations

This feature has no known activations.