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

 ACTIONS

-0.73

nep

-0.60

 mounts

-0.57

 planetary

-0.57

 stabilization

-0.57

 Jinn

-0.57

dan

-0.56

bows

-0.56

Pry

-0.56

flight

-0.55

POSITIVE LOGITS

ateful

0.75

lete

0.67

cigarettes

0.66

acist

0.66

amacare

0.65

ntil

0.63

icum

0.62

Semitism

0.62

onsequ

0.61

©¶æ¥µ

0.61

Activations Density 0.000%

No Known Activations

This feature has no known activations.