INDEX

Explanations

terms related to destruction and killing

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GEMMA-2-2B @ 13-gemmascope-res-65k

Configuration

google/gemma-scope-2b-pt-res/layer_13/width_65k/average_l0_75

Prompts (Dashboard)

24,576 prompts, 128 tokens each

Dataset (Dashboard)

monology/pile-uncopyrighted

Features

65,536

Data Type

float32

Hook Name

blocks.13.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

RunWith

-0.46

}{@

-0.46

 mergeFrom

-0.45

유

-0.44

 reag

-0.42

 kurtka

-0.42

 gawas

-0.41

beu

-0.41

 dàng

-0.41

tvguidetime

-0.40

POSITIVE LOGITS

 destroy

0.78

 destroying

0.77

 destroyed

0.74

 destroys

0.73

rmtree

0.71

掉的

0.67

 препратки

0.67

 destru

0.66

 destruir

0.66

DialogContent

0.65

Activations Density 0.335%

terms related to destruction and killing

No Comments

No Known Activations