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

iken

-0.82

 Reconstruction

-0.73

ãģĹ

-0.66

 Chao

-0.65

 Mong

-0.63

 malls

-0.61

ivism

-0.61

/*

-0.59

 Castle

-0.58

galitarian

-0.57

POSITIVE LOGITS

etheless

0.86

die

0.86

theless

0.81

edin

0.77

srf

0.74

 endif

0.73

ailable

0.72

anwhile

0.69

00200000

0.69

ensibly

0.68

Activations Density 0.000%

No Known Activations

This feature has no known activations.