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

ockey

-0.73

 Fault

-0.69

 Lodge

-0.68

 Manitoba

-0.67

 Hole

-0.66

 Winnipeg

-0.65

phis

-0.63

aku

-0.63

ierre

-0.63

 debtor

-0.62

POSITIVE LOGITS

 favorites

0.77

 favorite

0.69

alities

0.65

favorite

0.64

liest

0.62

 dexter

0.61

abouts

0.61

adj

0.60

ivas

0.60

 Accessed

0.60

Activations Density 0.000%

No Known Activations

This feature has no known activations.