INDEX

Explanations

No Explanations Found

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 3-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.3.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.3.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

zzle

-0.92

atta

-0.78

etz

-0.78

cott

-0.76

ivo

-0.75

Originally

-0.74

umb

-0.74

casters

-0.73

rea

-0.72

inker

-0.72

POSITIVE LOGITS

acles

0.70

 helper

0.69

 Recover

0.66

 Samar

0.65

 prospects

0.65

ãĥ¼ãĥĨ

0.64

onnaissance

0.64

ALSE

0.63

 defenses

0.62

 Landing

0.61

Activations Density 0.000%

No Known Activations

This feature has no known activations.