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

PlotsExplanationShow Test FieldDefault Test Text

IFrame

Link

Not in Any Lists

No Comments

Negative Logits

enzie

-0.80

uesday

-0.78

oeuv

-0.75

oward

-0.72

onduct

-0.71

 prelim

-0.70

urst

-0.70

psey

-0.69

cot

-0.68

illian

-0.67

POSITIVE LOGITS

Ish

0.68

izen

0.67

Atl

0.67

 Volt

0.67

 quake

0.65

Sense

0.65

 generation

0.65

erva

0.64

Yug

0.64

izens

0.63

Activations Density 0.000%

No Known Activations

This feature has no known activations.