INDEX

Explanations

programming elements, particularly related to function definitions and console logging

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

questions and code snippets

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

 informée

-0.69

 ब्रेकडाउन

-0.66

uxxxx

-0.65

Personendaten

-0.64

zzleHttp

-0.59

 surla

-0.59

tagHelperRunner

-0.59

 Alembic

-0.54

IsMutable

-0.54

aarrggbb

-0.53

POSITIVE LOGITS

 proprement

0.40

 espé

0.40

 profeta

0.39

 nôtre

0.38

 royaume

0.38

 Naciones

0.38

<eos>

0.38

 pertenece

0.38

 contesta

0.37

 thérape

0.37

Activations Density 0.011%

programming elements, particularly related to function definitions and console logging

questions and code snippets

No Comments

No Known Activations