INDEX

Explanations

elements related to structured response generation in an API context

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

programming structures

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GEMMA-2-9B-IT @ 20-gemmascope-res-131k

Configuration

google/gemma-scope-9b-it-res/layer_20/width_131k/average_l0_81

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

IFrame

Link

Not in Any Lists

No Comments

Negative Logits

 ComVisible

-0.48

 convertView

-0.47

Auswahl

-0.47

bumptech

-0.46

 kaynağından

-0.45

tus

-0.43

ped

-0.43

BagConstraints

-0.43

TokenNameLPAREN

-0.43

InputTagHelper

-0.42

POSITIVE LOGITS

verwijspagina

0.59

MLLoader

0.48

 surla

0.46

MessageState

0.44

MessageOf

0.41

QMetaType

0.40

lgari

0.39

centiles

0.39

parsedMessage

0.39

 Reports

0.37

Activations Density 0.027%

elements related to structured response generation in an API context

programming structures

No Comments

No Known Activations