INDEX

Explanations

structural patterns in code or markup language

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

copyright and include guards

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

 geſch

-1.05

 Geſch

-1.00

tagHelperRunner

-1.00

expandindo

-0.99

нгред

-0.97

AndEndTag

-0.96

 ویکی‌پدی

-0.94

 Geiſt

-0.93

 queſta

-0.92

majánló

-0.91

POSITIVE LOGITS

0.41

↵↵

0.41

0.40

0.39

Le

0.39

0.38

0.37

0.36

Le

0.36

Activations Density 0.029%

structural patterns in code or markup language

copyright and include guards

No Comments

No Known Activations