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

forward

-0.80

eri

-0.72

Ħ¢

-0.71

 merits

-0.70

aux

-0.68

 airs

-0.64

yi

-0.64

animous

-0.63

 constitu

-0.62

 prelim

-0.62

POSITIVE LOGITS

 followed

1.07

 about

0.82

natureconservancy

0.80

iPhone

0.69

igslist

0.68

BI

0.67

perties

0.65

about

0.64

Â·Â·

0.64

rounder

0.63

Activations Density 0.000%

No Known Activations

This feature has no known activations.