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

 Huss

-0.63

;;;;;;;;;;;;

-0.61

 Luxem

-0.61

 Worker

-0.60

 Subst

-0.60

 horizont

-0.60

Newsletter

-0.60

 Alger

-0.59

 Formation

-0.59

 Territ

-0.56

POSITIVE LOGITS

aylor

0.91

illion

0.80

ronic

0.71

lock

0.71

utra

0.70

icka

0.69

ocket

0.68

ishop

0.68

agos

0.67

entially

0.67

Activations Density 0.000%

No Known Activations

This feature has no known activations.