INDEX

Explanations

No Explanations Found

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 3-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.3.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.3.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

 Bonnie

-0.70

 Deborah

-0.68

 Aval

-0.67

 Norton

-0.64

opathy

-0.62

 Salv

-0.62

McH

-0.61

 Mast

-0.59

Whit

-0.59

 Bernstein

-0.59

POSITIVE LOGITS

abouts

0.79

rarily

0.76

gage

0.75

paralle

0.74

 atmosp

0.72

verty

0.68

Ì¶

0.67

pse

0.67

âĢ¦]

0.66

irements

0.65

Activations Density 0.000%

No Known Activations

This feature has no known activations.