INDEX

Explanations

special characters and formatting in code or text

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

plt.xticks

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

Lycka

-0.37

 aanwezig

-0.35

isPresent

-0.34

 auroit

-0.34

 Secretaría

-0.33

OrNull

-0.33

näm

-0.33

 Referential

-0.32

 ilustración

-0.31

 Comissão

-0.31

POSITIVE LOGITS

 propOrder

0.63

ArrowToggle

0.61

TypedDataSet

0.51

 <=",

0.51

domada

0.49

Bibliograf

0.49

Espèce

0.48

忑

0.48

IndentedString

0.47

 snippetHide

0.47

Activations Density 0.067%

special characters and formatting in code or text

plt.xticks

No Comments

No Known Activations

special characters and formatting in code or text

plt.xticks

No Comments

No Known Activations