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

IFrame

Link

Not in Any Lists

No Comments

Negative Logits

ĸļ

-0.87

¬¼

-0.77

cffff

-0.74

MpServer

-0.67

cci

-0.64

 dressing

-0.64

romeda

-0.64

 theoret

-0.64

ijk

-0.64

ascript

-0.63

POSITIVE LOGITS

adia

0.76

å¾

0.74

BLE

0.65

MORE

0.64

 refuted

0.63

ÙĴ

0.62

 demol

0.62

halla

0.61

 reminded

0.59

 Able

0.59

Activations Density 0.000%

No Known Activations

This feature has no known activations.