INDEX

Explanations

action words and expressions related to intent and purpose

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

following a phrase

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

帖最后由

-0.57

 disambiguazione

-0.57

 étoit

-0.56

 Theſe

-0.56

 dAtA

-0.56

WriteTagHelper

-0.55

.*")]

-0.54

 becauſe

-0.52

 nonUne

-0.52

ConstraintMaker

-0.52

POSITIVE LOGITS

nvarchar

0.38

ราย

0.37

pgterms

0.35

mitives

0.34

amssymb

0.34

씀

0.33

onnaissance

0.33

throp

0.32

 gynhyrchwyd

0.32

いく

0.31

Activations Density 0.042%

action words and expressions related to intent and purpose

following a phrase

No Comments

No Known Activations

action words and expressions related to intent and purpose

following a phrase

No Comments

No Known Activations