INDEX

Explanations

conjunctive phrases that indicate causality or conclusions

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

conjunctions of consequence

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

ReusableCell

-0.58

PreferredItem

-0.53

 ModelRenderer

-0.50

 Multiplex

-0.45

↹

-0.43

MockMvc

-0.43

entitled

-0.43

EqualTo

-0.43

بسم

-0.43

IUrlHelper

-0.42

POSITIVE LOGITS

ệc

0.45

Décès

0.42

SuppressMessage

0.41

 pomy

0.40

 DBNull

0.38

 Geduld

0.38

жнему

0.37

 isKindOfClass

0.37

 scio

0.36

 asumir

0.36

Activations Density 0.025%

conjunctive phrases that indicate causality or conclusions

conjunctions of consequence

No Comments

No Known Activations