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

 Compos

-0.71

 Scythe

-0.69

athe

-0.66

 Knot

-0.65

NCT

-0.65

 Astron

-0.64

Ear

-0.64

 Replay

-0.64

 Recorded

-0.64

 Immunity

-0.63

POSITIVE LOGITS

soon

0.84

oxide

0.70

Cruz

0.68

she

0.68

fuck

0.67

Ã±a

0.67

ãĥ¼ãĥ

0.67

 settlers

0.66

Ev

0.65

Eva

0.64

Activations Density 0.000%

No Known Activations

This feature has no known activations.