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

WAYS

-0.70

tti

-0.68

Untitled

-0.67

 Cosponsors

-0.64

mobi

-0.63

 overnight

-0.60

perm

-0.60

tera

-0.60

 underscore

-0.59

eteenth

-0.59

POSITIVE LOGITS

he

2.04

heit

1.04

hem

0.84

hei

0.77

htaking

0.75

heng

0.75

heet

0.71

she

0.70

ÃŃn

0.68

hed

0.68

Activations Density 0.000%

No Known Activations

This feature has no known activations.