INDEX

Explanations

references to technical specifications or details often used in descriptions of inventions or patents

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

technical configuration and code structure

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

 betweenstory

-0.81



-0.49

miyorum

-0.47

MessageOf

-0.46

StructEnd

-0.46

envolvimento

-0.44

 kaarangay

-0.44

íslu

-0.44

cessite

-0.44

 armée

-0.43

POSITIVE LOGITS

ArgumentParser

0.46

RTLR

0.42

NameInMap

0.42

GroupLayout

0.40

RB

0.39

 بيها

0.39

 חיצוניים

0.38

 fuller

0.38

MEMORANDUM

0.38

CreateModel

0.38

Activations Density 0.093%

references to technical specifications or details often used in descriptions of inventions or patents

technical configuration and code structure

No Comments

No Known Activations