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

£ı

-0.79

 directories

-0.67

 merchants

-0.67

 dism

-0.66

 creditors

-0.66

 Vendor

-0.65

 routing

-0.64

 descript

-0.64

 collector

-0.62

 scaven

-0.61

POSITIVE LOGITS

liest

0.81

OME

0.81

CLE

0.78

lier

0.78

bryce

0.75

gur

0.71

Tube

0.70

uay

0.70

hiba

0.69

AAF

0.69

Activations Density 0.000%

No Known Activations

This feature has no known activations.