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

adder

-0.76

 Antar

-0.75

 harb

-0.73

ãĤ©

-0.73

 Archdemon

-0.73

Dud

-0.72

igating

-0.71

look

-0.69

chwitz

-0.69

foundland

-0.68

POSITIVE LOGITS

 grain

0.66

 nutrition

0.65

ULAR

0.64

ividual

0.64

 complementary

0.64

ranch

0.62

 distribution

0.61

 perf

0.61

 utility

0.61

 crunch

0.61

Activations Density 0.000%

No Known Activations

This feature has no known activations.