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

PlotsExplanationShow Test FieldDefault Test Text

IFrame

Link

Not in Any Lists

No Comments

Negative Logits

 Qiao

-0.77

las

-0.69

Bei

-0.65

gars

-0.65

 Tanks

-0.64

laus

-0.63

apter

-0.60

olars

-0.59

 gigg

-0.59

avery

-0.58

POSITIVE LOGITS

 contingency

0.74

":[{"

0.72

ãĥ´ãĤ¡

0.71

unction

0.70

ãĥĥãĥī

0.70

jriwal

0.69

 è£ıç

0.67

Sche

0.66

activity

0.65

enza

0.63

Activations Density 0.000%

No Known Activations

This feature has no known activations.