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

hent

-0.90

afety

-0.80

xus

-0.78

mx

-0.75

ulhu

-0.74

kson

-0.71

cyclop

-0.71

itaire

-0.71

irgin

-0.70

ylum

-0.70

POSITIVE LOGITS

 once

0.87

 whiff

0.74

]).

0.69

 tongues

0.63

]),

0.61

 Flavoring

0.60

 Earn

0.59

]);

0.58

 Cold

0.58

 contag

0.58

Activations Density 0.000%

No Known Activations

This feature has no known activations.