INDEX

Explanations

No Explanations Found

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 1-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.1.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.1.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

 msec

-0.71

à©

-0.67

isan

-0.65

 Hort

-0.64

 inst

-0.62

"]=>

-0.61

CHR

-0.61

 clauses

-0.61

iciary

-0.61

cler

-0.60

POSITIVE LOGITS

ozy

0.71

ĸļ

0.69

 largeDownload

0.68

ovsky

0.67

İĭ

0.67

eness

0.67

Hav

0.66

oshop

0.65

 challeng

0.63

anooga

0.62

Activations Density 0.000%

No Known Activations

This feature has no known activations.