INDEX

Explanations

No Explanations Found

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 1-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.1.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.1.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

resent

-0.78

 newsp

-0.76

weights

-0.74

MRI

-0.70

animous

-0.68

 antib

-0.67

ieving

-0.66

hello

-0.65

untarily

-0.64

orsi

-0.64

POSITIVE LOGITS

 Cree

0.67

anders

0.63

ophen

0.62

 Trail

0.61

 Creek

0.61

 Airways

0.60

actionDate

0.59

lur

0.59

atche

0.58

 Spur

0.58

Activations Density 0.000%

No Known Activations

This feature has no known activations.