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

PlotsExplanationShow Test FieldDefault Test Text

IFrame

Link

Not in Any Lists

No Comments

Negative Logits

ham

-0.71

hao

-0.71

ulz

-0.71

gon

-0.64

 dunno

-0.63

gur

-0.63

Pik

-0.62

uay

-0.62

yon

-0.61

atar

-0.60

POSITIVE LOGITS

âĹ¼

0.74

phasis

0.72

 Citation

0.69

terness

0.68

--+

0.67

verages

0.66

benefit

0.66

ãĤ¼ãĤ¦ãĤ¹

0.65

edIn

0.65

ertodd

0.65

Activations Density 0.000%

No Known Activations

This feature has no known activations.