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

ĸļ

-0.72

ociate

-0.72

 ISBN

-0.70

 Scientist

-0.65

 Lloyd

-0.64

 caster

-0.62

 payout

-0.60

 Salmon

-0.59

 Kafka

-0.59

 WARN

-0.58

POSITIVE LOGITS

ENTION

0.76

riots

0.73

fighting

0.72

iet

0.67

================================================================

0.66

lessly

0.65

¶

0.65

½

0.65

bell

0.64

anamo

0.63

Activations Density 0.000%

No Known Activations

This feature has no known activations.