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

 paran

-0.74

 "$:/

-0.65

iannopoulos

-0.62

 Swim

-0.61

 nodd

-0.60

Cub

-0.59

oros

-0.59

ano

-0.59

rament

-0.59

 anomal

-0.58

POSITIVE LOGITS

sed

0.72

VK

0.70

ploy

0.68

dayName

0.66

starter

0.66

stood

0.66

iHUD

0.65

warts

0.65

INST

0.65

oglu

0.63

Activations Density 0.000%

No Known Activations

This feature has no known activations.