INDEX

Explanations

bracketed expressions or structured data formats

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

object initialization

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

-0.70

 eſt

-0.68

ſelves

-0.67

 ſou

-0.61

 ſtre

-0.61

 queſta

-0.61

 ſche

-0.60

 ſen

-0.59

 ſtate

-0.57

 pleaſure

-0.57

POSITIVE LOGITS

fjspx

0.60

ținut

0.53

nonatomic

0.50

ArgsConstructor

0.47

 constructing

0.45

</em>

0.44

 localObject

0.44

 constructed

0.44

 wholly

0.43

 clone

0.42

Activations Density 0.002%

bracketed expressions or structured data formats

object initialization

No Comments

No Known Activations