INDEX

Explanations

structures related to regions and associated codes

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

code colon

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

 мәкал

-0.94

 Мексичка

-0.91

 queſta

-0.89

 للمعارف

-0.88

تقاوى

-0.85



-0.84

########.

-0.83

WriteTagHelper

-0.82

TagMode

-0.81

 cherchés

-0.79

POSITIVE LOGITS

0.43

is

0.41

0.34

0.33

is

0.33

0.32

 either

0.32

0.31

↵

0.31

Activations Density 0.147%

structures related to regions and associated codes

code colon

No Comments

No Known Activations