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

 Ripple

-0.78

MpServer

-0.77

zbollah

-0.77

enegger

-0.75

 Rothschild

-0.70

 Rockefeller

-0.70

peria

-0.69

kefeller

-0.69

 sqor

-0.68

clinton

-0.67

POSITIVE LOGITS

attr

0.76

UV

0.74

 binds

0.73

 undisclosed

0.69

 searches

0.68

 search

0.67

////////////////

0.67

afe

0.66

 habitable

0.66

oki

0.62

Activations Density 0.000%

No Known Activations

This feature has no known activations.