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

ascript

-0.79

ģ«

-0.69

 Argon

-0.69

 Sass

-0.68

Za

-0.67

anguage

-0.67

Sok

-0.67

Translation

-0.65

åĤ

-0.63

 Metatron

-0.62

POSITIVE LOGITS

Pwr

0.75

cot

0.72

nis

0.68

psc

0.67

obil

0.66

 sidx

0.65

USD

0.65

letico

0.65

ruce

0.65

kee

0.64

Activations Density 0.000%

No Known Activations

This feature has no known activations.