INDEX

Explanations

patterns or structures in data represented by numerical values

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

 integrity

-0.38

 published

-0.37

 пора

-0.35

 protocol

-0.35

ˏ

-0.35

palindrome

-0.35

 InputDecoration

-0.35

/*---

-0.35

 valued

-0.34

preco

-0.34

POSITIVE LOGITS

 EconPapers

0.71

StoryboardSegue

0.57

BeginContext

0.57

addContainerGap

0.56

fjspx

0.54

DoubleQuotes

0.54

 autorytatywna

0.54

 propOrder

0.54

AndEndTag

0.51

 Theſe

0.47

Activations Density 0.061%

patterns or structures in data represented by numerical values

No Comments

No Known Activations