INDEX

Explanations

words related to endings or conclusions

oai_token-act-pair · gpt-3.5-turbo

instances of the word "end" or related forms

oai_token-act-pair · gpt-4o-mini Triggered by @bot

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

 blankets

-0.64

ãģ®ç

-0.62

 diapers

-0.61

 bulls

-0.60

 Factors

-0.60

sim

-0.60

 blanket

-0.59

 Boards

-0.58

 scoop

-0.58

 bubbles

-0.55

POSITIVE LOGITS

end

4.25

ends

2.94

END

2.80

ended

2.40

ending

2.33

enders

2.13

ender

2.01

endo

2.01

End

1.99

ende

1.83

Activations Density 0.013%

words related to endings or conclusions

instances of the word "end" or related forms

No Comments

No Known Activations