INDEX

Explanations

No Explanations Found

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 1-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.1.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.1.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

MpServer

-0.77

 Beir

-0.72

omin

-0.70

senal

-0.68

 Piper

-0.66

Orig

-0.66

 ceases

-0.66

SIGN

-0.64

awaru

-0.64

 qualify

-0.64

POSITIVE LOGITS

vern

0.73

doi

0.70

rying

0.66

vale

0.65

 Horton

0.65

liness

0.63

croft

0.62

furt

0.61

gregation

0.60

Activations Density 0.000%

No Known Activations

This feature has no known activations.