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

PlotsExplanationShow Test FieldDefault Test Text

IFrame

Link

Not in Any Lists

No Comments

Negative Logits

ystem

-0.71

OSH

-0.68

byter

-0.68

draft

-0.64

 recognition

-0.63

 admission

-0.60

abol

-0.59

 bracket

-0.58

Mew

-0.58

arer

-0.57

POSITIVE LOGITS

't

1.37

tein

0.82

ILLE

0.77

nos

0.72

emis

0.69

ioned

0.67

rouse

0.67

hig

0.67

ned

0.65

skirts

0.64

Activations Density 0.000%

No Known Activations

This feature has no known activations.