INDEX

Explanations

specific coding and programmatic terms

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

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

PlotsExplanationShow Test FieldDefault Test Text

IFrame

Link

Not in Any Lists

No Comments

Negative Logits

 Hitam

-0.42

toBeTruthy

-0.37

 betweenstory

-0.36

ReadLine

-0.36

 Perla

-0.35

 приве

-0.34

/[

-0.34

peach

-0.34

AutoField

-0.33

divarius

-0.33

POSITIVE LOGITS

 CreateTagHelper

0.89

Controls

0.81

SequentialGroup

0.63

 addSubview

0.61

tagHelperRunner

0.60

rawDesc

0.59

ScopeManager

0.56

RTHOOK

0.55

ftagPool

0.54

AnchorStyles

0.54

Activations Density 0.001%

specific coding and programmatic terms

No Comments

No Known Activations