INDEX

Explanations

No Explanations Found

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 5-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.5.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.5.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

 yourselves

-0.73

 orig

-0.66

 initiation

-0.66

 playbook

-0.66

TG

-0.65

 unden

-0.64

 alumni

-0.63

ãĥĻ

-0.58

TAM

-0.58

>)

-0.57

POSITIVE LOGITS

ened

0.93

enson

0.72

 Puzz

0.71

Ã©r

0.70

ector

0.70

ening

0.68

ersed

0.67

erion

0.67

aver

0.66

hair

0.65

Activations Density 0.000%

No Known Activations

This feature has no known activations.