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

PlotsExplanationShow Test FieldDefault Test Text

IFrame

Link

Not in Any Lists

No Comments

Negative Logits

illus

-0.80

isner

-0.68

 exclus

-0.66

ecycle

-0.65

 unsus

-0.63

merce

-0.62

 retali

-0.61

anded

-0.61

 upkeep

-0.61

 heroic

-0.60

POSITIVE LOGITS

posts

0.94

picture

0.74

Morning

0.72

Kar

0.70

kit

0.70

 Crescent

0.69

izon

0.69

ÙĬ

0.69

ãĥķãĤ©

0.68

eral

0.68

Activations Density 0.000%

No Known Activations

This feature has no known activations.