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

uyomi

-0.79

£ı

-0.77

ibaba

-0.76

onge

-0.72

ĵĺ

-0.72

ĸļ

-0.71

igslist

-0.70

 consum

-0.68

xtap

-0.67

 WATCHED

-0.67

POSITIVE LOGITS

ALP

0.83

HQ

0.70

NIC

0.69

PA

0.68

fell

0.66

 hots

0.64

African

0.64

>>>>>>>>

0.63

GC

0.63

NP

0.63

Activations Density 0.000%

No Known Activations

This feature has no known activations.