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

catch

-0.75

":""},{"

-0.72

suits

-0.71

âĸ¬âĸ¬

-0.69

lap

-0.69

dating

-0.68

cock

-0.67

 Cele

-0.64

 Race

-0.64

 kisses

-0.63

POSITIVE LOGITS

ithing

0.84

ascus

0.79

monary

0.78

 externalToEVAOnly

0.74

orah

0.70

negie

0.70

Desk

0.69

 upkeep

0.69

undown

0.69

udging

0.62

Activations Density 0.000%

No Known Activations

This feature has no known activations.