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

SHIP

-0.79

 Annotations

-0.79

script

-0.71

Appearance

-0.70

":""},{"

-0.64

Script

-0.63

attered

-0.62

issa

-0.61

hend

-0.61

 unmarked

-0.60

POSITIVE LOGITS

phabet

0.95

 deregulation

0.83

 regulators

0.72

oppable

0.71

ucle

0.69

 Resist

0.67

usters

0.67

EStreamFrame

0.66

 resist

0.66

acan

0.65

Activations Density 0.000%

No Known Activations

This feature has no known activations.