INDEX

Explanations

No Explanations Found

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 3-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.3.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.3.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

 dehuman

-0.72

agonists

-0.71

 largeDownload

-0.69

pez

-0.68

capt

-0.67

wolf

-0.67

 Skydragon

-0.66

 scapego

-0.66

 emot

-0.66

ority

-0.65

POSITIVE LOGITS

iche

0.69

 Harbour

0.64

 gravel

0.63

 Shant

0.63

rench

0.63

ortunately

0.63

ãĥ³ãĤ¸

0.63

 Bangalore

0.63

ãĤ®

0.62

 Hastings

0.61

Activations Density 0.000%

No Known Activations

This feature has no known activations.