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

uration

-0.81

NG

-0.75

ystem

-0.73

 mosqu

-0.73

sylv

-0.72

sil

-0.70

 wrench

-0.69

resist

-0.68

stru

-0.68

STER

-0.67

POSITIVE LOGITS

Coh

0.88

 Schwar

0.72

 Gaul

0.71

Ake

0.70

 Carth

0.69

 Citation

0.66

 quotations

0.66

Sa

0.66

Bav

0.66

 Piano

0.65

Activations Density 0.000%

No Known Activations

This feature has no known activations.