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

emetery

-0.75

 Zheng

-0.72

 famine

-0.65

Eco

-0.65

 rehearsal

-0.62

 Travels

-0.62

ihad

-0.62

Mum

-0.62

 Wool

-0.61

 Cornel

-0.60

POSITIVE LOGITS

properties

0.76

ĪĴ

0.74

 denomin

0.69

pport

0.67

generic

0.66

mask

0.66

MID

0.65

phase

0.65

layer

0.65

resources

0.65

Activations Density 0.000%

No Known Activations

This feature has no known activations.