INDEX

Explanations

specific numerical values or counts related to data, measurements, or statistics

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

technical documentation and naming

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

 Kirsch

-0.42

 Brill

-0.41

 aparte

-0.41

asztok

-0.40

OrBuilder

-0.39

 sencillos

-0.39

plit

-0.38

getJSONArray

-0.38

 Scene

-0.37

 useHistory

-0.37

POSITIVE LOGITS

 للاسماء

0.58

 dAtA

0.55

 ComVisible

0.54

AndEndTag

0.51

ValueStyle

0.47

 typelib

0.45



0.44

OGND

0.41

httphttps

0.41

DispatchToProps

0.41

Activations Density 0.037%

specific numerical values or counts related to data, measurements, or statistics

technical documentation and naming

No Comments

No Known Activations