INDEX

Explanations

No Explanations Found

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 3-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.3.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.3.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

otypes

-0.66

 Hind

-0.66

roups

-0.65

vernment

-0.65

ieties

-0.65

PKK

-0.65

redo

-0.65

ogen

-0.64

ird

-0.63

 Mongolia

-0.62

POSITIVE LOGITS

ealous

0.77

REDACTED

0.76

 Examiner

0.72

PLA

0.71

Consumer

0.66

upon

0.64

bid

0.64

asonic

0.64

omsky

0.63

Downloadha

0.63

Activations Density 0.000%

No Known Activations

This feature has no known activations.