INDEX

Explanations

programming constructs and error handling mechanisms

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

programming keywords and constructs

np_acts-logits-general · gemini-2.5-flash-lite

New Auto-Interp

Configuration

google/gemma-scope-9b-pt-res/layer_24/width_131k/average_l0_97

Prompts (Dashboard)

24,576 prompts, 128 tokens each

Dataset (Dashboard)

monology/pile-uncopyrighted

Features

131,072

Data Type

float32

Hook Name

blocks.24.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

NOU

-0.45

 getattr

-0.44

Cry

-0.43

 Tern

-0.39

ftagPool

-0.39

cry

-0.39

</em>

-0.39

примеча

-0.39

EU

-0.38

evsky

-0.38

POSITIVE LOGITS

 AssemblyTitle

0.70

sizeCache

0.60

enterOuterAlt

0.55

帖最后由

0.54

 useAppContext

0.53

 typelib

0.49

 ModelExpression

0.48

 виправивши

0.47

 tartalomajánló

0.47

AddHtmlAttribute

0.47

Activations Density 0.030%

programming constructs and error handling mechanisms

programming keywords and constructs

No Comments

No Known Activations