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

[_

-0.78

ãĥ¼ãĥĨãĤ£

-0.77

ãĥķãĤ¡

-0.73

ãĥ¼ãĥ³

-0.71

ODY

-0.69

matched

-0.69

ãĤ©

-0.66

Raven

-0.66

 sugars

-0.66

Â¢

-0.65

POSITIVE LOGITS

 respect

0.80

 landslide

0.74

ratulations

0.66

anmar

0.63

ilion

0.62

 Cartoon

0.61

cule

0.60

gnu

0.59

 inund

0.59

eco

0.58

Activations Density 0.000%

No Known Activations

This feature has no known activations.