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

 Citiz

-0.78

\\\\\\\\\\\\\\\\

-0.73

 eleph

-0.73

################################

-0.71

CHO

-0.65

********************************

-0.64

 derives

-0.61

 à¤

-0.60

--------------------------------

-0.59

 Autism

-0.58

POSITIVE LOGITS

olson

0.90

regor

0.86

regation

0.79

regate

0.78

ensional

0.73

ĸļ

0.71

 Warden

0.70

urated

0.68

urable

0.68

herty

0.67

Activations Density 0.000%

No Known Activations

This feature has no known activations.