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

 Socket

-0.74

breaker

-0.71

istry

-0.71

lessly

-0.68

 Elixir

-0.67

SAP

-0.66

strom

-0.65

GHz

-0.63

ancers

-0.63

 Loading

-0.63

POSITIVE LOGITS

 negro

0.72

egal

0.71

alach

0.70

rontal

0.68

Latin

0.66

erguson

0.66

rab

0.65

raltar

0.65

aucas

0.65

 kosher

0.65

Activations Density 0.000%

No Known Activations

This feature has no known activations.