INDEX

Explanations

specific coding symbols and terms related to programming logic

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

absorption bands

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

 informée

-0.72

########.

-0.68

GEBURTSDATUM

-0.60

EDEFAULT

-0.59

Tembelea

-0.59

 nakalista

-0.57

 ujednoznacz

-0.57

principalColumn

-0.56

 мәкал

-0.56

WriteTagHelper

-0.55

POSITIVE LOGITS

rubin

0.34

Dez

0.32

rog

0.32

"{

0.32

 Sheraton

0.31

 dehors

0.30

})`

0.30

 pear

0.29

 fles

0.29

 iodine

0.29

Activations Density 0.037%

specific coding symbols and terms related to programming logic

absorption bands

No Comments

No Known Activations