INDEX

Explanations

occurrences of numerical values and their formatting in various contexts

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

`src/` paths

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

findpost

-0.50

Litteratur

-0.47

 ſind

-0.45

dafx

-0.43

 Motta

-0.41

Heb

-0.41

 urllib

-0.41

 leading

-0.41

MatIconModule

-0.41

 ſch

-0.40

POSITIVE LOGITS

Datuak

0.55

ResumeLayout

0.52

 informée

0.51

 bezeichneter

0.47

請繼續往下閱讀

0.47

enterOuterAlt

0.44

AutoScale

0.42

 NSCoder

0.42

IVEREF

0.41

 indisponible

0.41

Activations Density 0.002%

occurrences of numerical values and their formatting in various contexts

`src/` paths

No Comments

No Known Activations