INDEX

Explanations

No Explanations Found

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 4-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.4.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.4.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

oyle

-0.68

instead

-0.63

 combustion

-0.63

Otherwise

-0.62

bard

-0.62

Capital

-0.61

duc

-0.61

 interchange

-0.59

Raw

-0.59

hani

-0.57

POSITIVE LOGITS

Zin

0.76

ibaba

0.72

 Samoa

0.71

Ys

0.67

assic

0.66

 Gleaming

0.66

miah

0.65

 Loren

0.64

thood

0.64

ĸļ

0.64

Activations Density 0.000%

No Known Activations

This feature has no known activations.