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

Not in Any Lists

No Comments

Negative Logits

 enrichment

-0.65

'';

-0.64

 Teachers

-0.63

 disclaimer

-0.60

 multiplier

-0.60

ENA

-0.60

-0.59

du

-0.59

 diploma

-0.59

 labour

-0.58

POSITIVE LOGITS

Blade

0.76

urbed

0.73

oid

0.71

rent

0.68

android

0.68

bilt

0.67

ilst

0.67

abilia

0.67

vr

0.67

Ð¾

0.67

Activations Density 0.000%

No Known Activations

This feature has no known activations.