INDEX

Explanations

programming code structures and syntax elements

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

user followed by data keys

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

 bezeichneter

-0.52

 consectetur

-0.36

 Вікіпе

-0.35

 Wikiseite

-0.35

CppCodeGen

-0.33

TemporalType

-0.32

flashdata

-0.32

 pinulongan

-0.32

цездатний

-0.32

تقاوى

-0.31

POSITIVE LOGITS

 queſto

0.65

andExpect

0.64

 Paglinawan

0.62

WireFormatLite

0.57

AndEndTag

0.56

 zwiſchen

0.56

 daysTop

0.55

⏡

0.54

[@BOS@]

0.54

<unused8>

0.54

Activations Density 0.026%

programming code structures and syntax elements

user followed by data keys

No Comments

No Known Activations