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

 Notting

-0.80

 Shinra

-0.73

pse

-0.70

 muster

-0.66

 methyl

-0.65

ãĥŀ

-0.65

uyomi

-0.63

Pu

-0.63

enance

-0.63

 nomine

-0.63

POSITIVE LOGITS

shots

0.79

reads

0.76

hari

0.70

hots

0.69

agles

0.68

âķĲ

0.68

fires

0.67

birds

0.66

rites

0.66

 knees

0.66

Activations Density 0.000%

No Known Activations

This feature has no known activations.