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

WB

-0.76

_-

-0.71

 Tycoon

-0.70

 Ruler

-0.68

MODE

-0.68

ding

-0.64

pmwiki

-0.64

Pwr

-0.60

Heavy

-0.59

rawler

-0.59

POSITIVE LOGITS

quet

0.75

Ã±o

0.74

 available

0.72

 dissemin

0.69

igible

0.69

ests

0.67

Downloadha

0.67

isi

0.67

ellen

0.66

 anonym

0.62

Activations Density 0.000%

No Known Activations

This feature has no known activations.