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

nt

-0.77

ailed

-0.73

sp

-0.72

fa

-0.71

bright

-0.69

ako

-0.69

vers

-0.67

ailing

-0.67

aker

-0.66

NJ

-0.66

POSITIVE LOGITS

 horizont

0.66

 condol

0.62

 commissions

0.62

 Frames

0.61

 glim

0.61

 murd

0.60

OOOOOOOO

0.60

 Ambro

0.60

 Reserv

0.59

 constitu

0.59

Activations Density 0.000%

No Known Activations

This feature has no known activations.