INDEX

Explanations

No Explanations Found

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 1-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.1.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.1.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

 Urug

-0.71

api

-0.71

Ups

-0.63

rep

-0.62

 north

-0.62

iverse

-0.62

oyal

-0.61

parts

-0.61

otted

-0.60

 Gupta

-0.60

POSITIVE LOGITS

uces

0.81

 sugg

0.79

akeru

0.73

emetery

0.70

XT

0.70

 mathemat

0.70

uesday

0.68

 accomp

0.68

[+

0.67

sein

0.67

Activations Density 0.000%

No Known Activations

This feature has no known activations.