INDEX

Explanations

technical terms related to programming and data handling

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

exceeding limits or ranges

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

bootstrapcdn

-0.63

 '\\;'

-0.60

帖最后由

-0.56

 ModelRenderer

-0.56

 dAtA

-0.55

+#+#

-0.51

rawDesc

-0.50

Tembelea

-0.50

 allAfrica

-0.49

ImageContext

-0.48

POSITIVE LOGITS

 achieved

0.47

 complied

0.46

 Abweich

0.44

 compliance

0.42

 cumplimiento

0.41

KEYCODE

0.40

 épaules

0.39

 Achie

0.39

 accomplished

0.38

Achie

0.38

Activations Density 0.130%

technical terms related to programming and data handling

exceeding limits or ranges

No Comments

No Known Activations