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

wrapper

-0.76

pour

-0.69

ora

-0.66

oslov

-0.64

chemy

-0.63

row

-0.62

bugs

-0.61

oire

-0.61

onna

-0.61

ft

-0.60

POSITIVE LOGITS

terday

0.86

ĸļ

0.73

#$

0.66

rx

0.64

 incent

0.63

 helicop

0.63

 guid

0.60

 Pascal

0.60

OMG

0.60

 Mens

0.59

Activations Density 0.000%

No Known Activations

This feature has no known activations.