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

ebus

-0.68

etheless

-0.68

deen

-0.67

wang

-0.66

gang

-0.62

';

-0.61

merga

-0.59

 Kurd

-0.58

fleet

-0.58

Erd

-0.58

POSITIVE LOGITS

anya

0.71

veyard

0.71

ourses

0.71

essions

0.70

aviour

0.68

ne

0.68

ierre

0.65

aving

0.65

ESSION

0.65

 Mellon

0.65

Activations Density 0.000%

No Known Activations

This feature has no known activations.