INDEX

Explanations

actions or instructions related to citing sources and navigating digital content

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

action verbs and complements

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

IBOutlet

-0.53

ViewFeatures

-0.44

 danke

-0.42

TemporalType

-0.42

PerformLayout

-0.41

Greetings

-0.40

 kasarigan

-0.40

 recuper

-0.39

client

-0.39

 위한

-0.39

POSITIVE LOGITS

 avoient

0.55

Hvis

0.46

ſelves

0.46

 Hvis

0.46

UseVisualStyle

0.45

ſelf

0.45

 ſeine

0.42

曖昧さ

0.42

的话

0.41

 étoient

0.41

Activations Density 0.067%

actions or instructions related to citing sources and navigating digital content

action verbs and complements

No Comments

No Known Activations