INDEX

Explanations

excessive or extreme behaviors and their potential negative impacts

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

 Protocol

-0.39

 للمعارف

-0.35

ені

-0.33

 xác

-0.32

madı

-0.32

iligung

-0.32

via

-0.31

stateMutability

-0.31

 Steck

-0.30

 acreditar

-0.30

POSITIVE LOGITS

awtextra

0.69

MemoryWarning

0.66

 ModelExpression

0.62

⤹

0.57

ſelves

0.56

 ainfi

0.56

 AssemblyCulture

0.54

VYMaps

0.54

 pleaſure

0.54

 faſt

0.53

Activations Density 0.060%

excessive or extreme behaviors and their potential negative impacts

No Comments

No Known Activations