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

HK

-0.79

 organise

-0.71

 Expend

-0.69

 Delhi

-0.67

eworks

-0.66

Boo

-0.65

 GOODMAN

-0.65

Inv

-0.64

Hold

-0.64

HOU

-0.64

POSITIVE LOGITS

 herald

0.65

 infall

0.64

voy

0.63

 proclaiming

0.62

 gunshot

0.61

cellent

0.61

gments

0.60

olo

0.60

 Semin

0.59

 gull

0.59

Activations Density 0.000%

No Known Activations

This feature has no known activations.