INDEX

Explanations

elements related to causal relationships and consequences

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

New Auto-Interp

Configuration

google/gemma-scope-9b-pt-res/layer_20/width_131k/average_l0_114

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

contentLoaded

-0.77

 betweenstory

-0.75

OGND

-0.73

 ویکی‌پدیا

-0.68

 насељу

-0.65

évaluateur

-0.65

makeText

-0.65

webElementGuid

-0.65

 فريبيس

-0.64

 للاسماء

-0.64

POSITIVE LOGITS

 directly

0.38

directly

0.34

 direct

0.31

 derived

0.31

grunn

0.31

而来

0.30

 plane

0.29

 langsung

0.29

 subsequent

0.28

 direkt

0.28

Activations Density 0.131%

elements related to causal relationships and consequences

No Comments

No Known Activations