INDEX

Explanations

patterns and data structures in scientific research outputs

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

figure references

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GEMMA-2-9B-IT @ 20-gemmascope-res-131k

Configuration

google/gemma-scope-9b-it-res/layer_20/width_131k/average_l0_81

Prompts (Dashboard)

24,576 prompts, 128 tokens each

Dataset (Dashboard)

monology/pile-uncopyrighted

Features

131,072

Data Type

float32

Hook Name

blocks.20.hook_resid_post

Hook Layer

Architecture

jumprelu

Context Size

1,024

Dataset

monology/pile-uncopyrighted

Activation Function

relu

Embeds

IFrame

Link

Not in Any Lists

No Comments

Negative Logits

Hentet

-0.78

 Monfieur

-0.77

queryInterface

-0.75

awtextra

-0.68

 ſind

-0.66

 canst

-0.66

OGND

-0.65

 resourceCulture

-0.65

AndEndTag

-0.65

httphttps

-0.65

POSITIVE LOGITS

 бывают

0.28

jsxFileName

0.28

dne

0.28

 bekannt

0.28

 Join

0.27

bek

0.27

 rencontres

0.27

SetActive

0.26

HomeAsUp

0.26

Join

0.26

Activations Density 0.039%

patterns and data structures in scientific research outputs

figure references

No Comments

No Known Activations