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

=-=-=-=-

-0.84

Reviewer

-0.80

 Leilan

-0.78

arial

-0.77

\\\\\\\\

-0.76

SHIP

-0.75

vous

-0.66

«ĺ

-0.64

 Skydragon

-0.64

xxxxxxxx

-0.63

POSITIVE LOGITS

aware

0.70

gins

0.70

got

0.69

igs

0.66

agh

0.66

 Spartans

0.65

rogram

0.65

 shutdown

0.63

aith

0.62

 Gren

0.61

Activations Density 0.000%

No Known Activations

This feature has no known activations.