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

Austral

-0.84

Magikarp

-0.76

Vs

-0.75

Episode

-0.71

Vert

-0.69

Reloaded

-0.69

Kay

-0.68

Land

-0.68

NW

-0.67

Percent

-0.66

POSITIVE LOGITS

blance

0.76

ibaba

0.76

assian

0.75

 fuse

0.69

daq

0.68

 Magicka

0.65

ilage

0.65

 drills

0.64

agher

0.64

vision

0.64

Activations Density 0.000%

No Known Activations

This feature has no known activations.