INDEX

Explanations

No Explanations Found

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 3-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.3.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.3.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

YA

-0.74

IVES

-0.72

âĢ¦]

-0.72

Sund

-0.72

ãĤ¦ãĤ¹

-0.71

IONS

-0.70

selves

-0.70

pload

-0.70

 throats

-0.70

ocular

-0.66

POSITIVE LOGITS

eton

0.73

 Kosovo

0.70

 Vermont

0.68

 undergrad

0.68

 Quake

0.66

 Prometheus

0.66

rero

0.65

RSA

0.65

 Burlington

0.64

 Buchanan

0.63

Activations Density 0.000%

No Known Activations

This feature has no known activations.