INDEX

Explanations

terms related to technical processes and systems

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

programming terms

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

 calitate

-0.44

 equipado

-0.41

 equipada

-0.39

ServletException

-0.37

layoutControl

-0.37

 echip

-0.37

 sólida

-0.34

 kapasitas

-0.34

 construida

-0.34

 acompañada

-0.34

POSITIVE LOGITS

WriteTagHelper

0.74

GEBURTSDATUM

0.66

PerformLayout

0.64

transQ

0.62

 autorytatywna

0.61

SequentialGroup

0.59

 jsPsych

0.59

ArgsConstructor

0.57

Tembelea

0.57

rungsseite

0.57

Activations Density 0.120%

terms related to technical processes and systems

programming terms

No Comments

No Known Activations