INDEX

Explanations

No Explanations Found

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 1-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.1.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.1.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

ppo

-0.69

may

-0.69

nce

-0.69

wm

-0.68

orem

-0.66

é¾

-0.65

ãĤ¯

-0.64

less

-0.64

 Doct

-0.63

pause

-0.63

POSITIVE LOGITS

iga

0.69

 Consulting

0.65

 vying

0.63

 blasting

0.61

 vice

0.61

aughs

0.60

 runaway

0.60

 bargaining

0.59

leigh

0.58

 Capital

0.58

Activations Density 0.000%

No Known Activations

This feature has no known activations.