INDEX

Explanations

code-related phrases indicating function definitions and actions within programming contexts

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

Sites, tests, descriptions

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

fromnode

-0.75

ConstraintMaker

-0.75

 jsPsych

-0.73

.*")]

-0.72

 AssemblyTitle

-0.70

AddTagHelper

-0.69

帖最后由

-0.68

+:+

-0.68

AnimationsModule

-0.65

UnusedPrivate

-0.65

POSITIVE LOGITS

 feroit

0.41

 enfans

0.36

 descrição

0.35

 pouvoit

0.35

 justiça

0.34

 mères

0.34

ḩ

0.34

 Kerajaan

0.33

 dévelo

0.32

Filmografie

0.32

Activations Density 0.005%

code-related phrases indicating function definitions and actions within programming contexts

Sites, tests, descriptions

No Comments

No Known Activations