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

kus

-0.79

stasy

-0.78

lesh

-0.75

othy

-0.72

piracy

-0.70

bris

-0.70

qqa

-0.69

imei

-0.69

 Territories

-0.69

phal

-0.69

POSITIVE LOGITS

idad

0.71

 dial

0.68

CLS

0.67

MEN

0.66

den

0.64

 Lowry

0.64

 AMER

0.62

 Athen

0.62

 hyster

0.62

DERR

0.61

Activations Density 0.000%

No Known Activations

This feature has no known activations.