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

selection

-0.73

 selection

-0.69

 Selection

-0.69

change

-0.67

 translator

-0.64

 translation

-0.64

advertisement

-0.63

aning

-0.62

 interpreter

-0.62

 Change

-0.62

POSITIVE LOGITS

ľ

1.21

£ı

0.96

©¶æ

0.94

»Ĵ

0.94

ĪĴ

0.91

©¶æ¥µ

0.87

 Archdemon

0.81

Ľ

0.77

Ĭ±

0.76

Mub

0.76

Activations Density 0.000%

No Known Activations

This feature has no known activations.