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

 Cron

-0.68

Cs

-0.67

 Pats

-0.66

ixties

-0.64

 Corporate

-0.62

HH

-0.62

 Clintons

-0.62

tails

-0.61

VB

-0.60

payer

-0.59

POSITIVE LOGITS

 livest

0.89

sembly

0.88

uria

0.86

vertisement

0.86

 glim

0.81

 seiz

0.79

miah

0.77

ateur

0.76

aque

0.75

ratulations

0.74

Activations Density 0.000%

No Known Activations

This feature has no known activations.