INDEX

Explanations

key phrases associated with causality and consequences

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

going, leading, doing

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

FunctionFlags

-0.49

makeConstraints

-0.49

Хьажоргаш

-0.47

OCCURRED

-0.46

 PartialEq

-0.45

tonsoft

-0.40

WithIOException

-0.39

انجليز

-0.38

gram

-0.38

ніципалі

-0.38

POSITIVE LOGITS

 שוליים

0.45

homonymie

0.40

 насеље

0.39

DockStyle

0.39

addPreferredGap

0.38

acherous

0.36



0.36

UserScript

0.36

ьаж

0.35

AnchorTagHelper

0.35

Activations Density 0.277%

key phrases associated with causality and consequences

going, leading, doing

No Comments

No Known Activations