INDEX

Explanations

terms related to machine learning datasets and handwritten digit recognition

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

handwritten characters or ImageNet dataset

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

Hochspringen

-0.70

__(/*!

-0.59

 صوتيه

-0.54

verwijspagina

-0.51

 kasarigan

-0.49

 propOrder

-0.47

Revenir

-0.47

SequentialGroup

-0.47

Skocz

-0.47

 שוליים

-0.44

POSITIVE LOGITS

mnist

0.90

sidemargin

0.59

 torchvision

0.47

 Normdatei

0.46

bootstrapcdn

0.44

fillType

0.43

 farfetch

0.42

 Minist

0.41

 getActivity

0.40

 bookId

0.40

Activations Density 0.007%

terms related to machine learning datasets and handwritten digit recognition

handwritten characters or ImageNet dataset

No Comments

No Known Activations