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

 Picks

-0.74

 Freak

-0.69

 Minion

-0.64

CD

-0.63

ipp

-0.61

 picks

-0.61

 Fantasy

-0.60

SJ

-0.60

Ped

-0.60

 Sonic

-0.59

POSITIVE LOGITS

 appre

0.85

 redes

0.82

milo

0.76

 volunt

0.75

roleum

0.70

lehem

0.65

 Conclusion

0.65

translation

0.65

unal

0.65

 Floating

0.64

Activations Density 0.000%

No Known Activations

This feature has no known activations.