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

²¾

-0.80

ocese

-0.72

onto

-0.71

Ħ¢

-0.70

uded

-0.70

otto

-0.69

Nim

-0.68

ennes

-0.66

hin

-0.66

erno

-0.66

POSITIVE LOGITS

 survival

1.45

 Survival

1.00

surv

0.67

reach

0.64

Untitled

0.63

merce

0.63

lined

0.63

EngineDebug

0.61

rawdownloadcloneembedreportprint

0.61

Recomm

0.61

Activations Density 0.000%

No Known Activations

This feature has no known activations.