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

 Airways

-0.85

ãĥķãĤ¡

-0.79

Tok

-0.75

Â·Â·

-0.75

FORE

-0.72

 Frankfurt

-0.72

Ku

-0.67

ngth

-0.67

 Citiz

-0.66

LIST

-0.63

POSITIVE LOGITS

irez

0.76

orously

0.74

ragon

0.71

yards

0.69

ranch

0.69

osed

0.68

grain

0.67

mobile

0.65

uces

0.64

ministic

0.63

Activations Density 0.000%

No Known Activations

This feature has no known activations.