INDEX

Explanations

phrases indicating limitations or conditions in actions or processes

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

cannot [verb]

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

 tend

-0.38

テンツ

-0.37

kaŭ

-0.35

SOUNDBITE

-0.34

SourceChecksum

-0.33

 accords

-0.33

jsxFileName

-0.32

 متعلقه

-0.31

 VIDEOT

-0.31

 spat

-0.30

POSITIVE LOGITS

 Cannot

0.81

Cannot

0.77

impossible

0.75

 unmöglich

0.74

cannot

0.72

 impossível

0.71

Impossible

0.70

 cannot

0.70

httphttps

0.66

 impossible

0.65

Activations Density 0.141%

phrases indicating limitations or conditions in actions or processes

cannot [verb]

No Comments

No Known Activations