INDEX

Explanations

quotation marks and references to notable articles or publications

oai_token-act-pair · gpt-4o-mini

Triggered by @bot

citations and references

np_acts-logits-general · gemini-2.5-flash-lite

New Auto-Interp

Configuration

google/gemma-scope-9b-pt-res/layer_24/width_131k/average_l0_97

Prompts (Dashboard)

24,576 prompts, 128 tokens each

Dataset (Dashboard)

monology/pile-uncopyrighted

Features

131,072

Data Type

float32

Hook Name

blocks.24.hook_resid_post

Hook Layer

Architecture

jumprelu

Context Size

1,024

Dataset

monology/pile-uncopyrighted

Activation Function

relu

Embeds

IFrame

Link

Negative Logits

Билгалдахарш

-0.41

ivant

-0.36

 break

-0.35

Anmerkungen

-0.34

ябре

-0.33

not

-0.32

 հղումներ

-0.32

ph

-0.31

 Conclusion

-0.31

POSITIVE LOGITS

 ResponseEntity

0.63

extAlignment

0.60

 nakalista

0.53

 ainfi

0.53

ValueStyle

0.50

硃

0.50

reportWebVitals

0.50

DockStyle

0.50

 パンチラ

0.50

 好文分享

0.49

Activations Density 0.018%

quotation marks and references to notable articles or publications

citations and references

No Known Activations