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

 Films

-0.74

eco

-0.67

alian

-0.65

Zoo

-0.64

 slaughter

-0.63

isine

-0.63

 Harvest

-0.62

Han

-0.61

vised

-0.60

ixtures

-0.60

POSITIVE LOGITS

/#

0.71

CLOSE

0.69

utterstock

0.69

``

0.67

DOM

0.66

TS

0.66

Connector

0.66

 REPL

0.61

Newsletter

0.61

 Contin

0.61

Activations Density 0.000%

No Known Activations

This feature has no known activations.