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

 referen

-0.80

ritch

-0.76

 unemploy

-0.75

BLE

-0.67

emetery

-0.66

anguage

-0.65

nai

-0.65

enough

-0.63

arten

-0.63

colored

-0.62

POSITIVE LOGITS

 really

0.87

 truly

0.80

 Corinth

0.75

 REALLY

0.74

 Tycoon

0.73

 Typh

0.68

 Leilan

0.67

UTC

0.67

 Mysteries

0.66

login

0.64

Activations Density 0.000%

No Known Activations

This feature has no known activations.