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

IFrame

Link

Not in Any Lists

No Comments

Negative Logits

adows

-0.71

lessly

-0.68

 fung

-0.66

aily

-0.63

isively

-0.63

 treasury

-0.62

 harm

-0.61

otrop

-0.60

 sleeves

-0.60

othing

-0.60

POSITIVE LOGITS

 followed

2.20

 preceded

1.39

 trailed

0.84

 accompanied

0.83

 supplemented

0.82

 WATCHED

0.80

roleum

0.78

 greeted

0.76

hower

0.75

 heard

0.74

Activations Density 0.000%

No Known Activations

This feature has no known activations.