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

alle

-0.76

 Â®

-0.66

":[

-0.64

MODE

-0.64

\":

-0.62

ITNESS

-0.62

ILS

-0.61

ï¿½ï¿½

-0.60

actionDate

-0.60

 remotely

-0.59

POSITIVE LOGITS

 Gibbs

0.84

iatus

0.78

igslist

0.70

 Qiao

0.68

 corpus

0.67

 roundup

0.64

neb

0.63

 Faust

0.62

 clipboard

0.62

 subp

0.62

Activations Density 0.000%

No Known Activations

This feature has no known activations.