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

 Cups

-0.69

 Hust

-0.66

iments

-0.65

 Voters

-0.63

Saf

-0.61

 comr

-0.61

 âĺ

-0.61

 Curt

-0.60

 metaphors

-0.60

 Bachelor

-0.60

POSITIVE LOGITS

eln

0.75

ondon

0.74

maxwell

0.73

ento

0.71

anian

0.68

clusive

0.67

naire

0.66

minster

0.66

rush

0.64

hyde

0.63

Activations Density 0.000%

No Known Activations

This feature has no known activations.