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

Negative Logits

 Languages

-0.74

agher

-0.74

erest

-0.73

CRC

-0.72

idth

-0.72

iffe

-0.71

jab

-0.69

roxy

-0.69

 IMAGES

-0.68

views

-0.66

POSITIVE LOGITS

 prevail

0.67

 outdoors

0.65

icky

0.65

 srfAttach

0.65

hor

0.65

 rall

0.65

Ĩ

0.65

 ratt

0.63

©¶æ

0.62

 oppos

0.62

Activations Density 0.000%

No Known Activations

This feature has no known activations.