INDEX

Explanations

references to openings and patterns in technical descriptions

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

holes and openings

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

 wezen

-0.48

 solteiro

-0.46

 føl

-0.45

 mesmas

-0.45

 engraçadas

-0.44

midler

-0.44

 zijne

-0.43

 vectoriales

-0.42

 completos

-0.42

 hayas

-0.41

POSITIVE LOGITS

 holes

0.95

 hole

0.91

 Holes

0.83

holes

0.82

 HOLE

0.81

 Hole

0.79

hole

0.78

Hole

0.77

 perforation

0.74

Holes

0.70

Activations Density 0.041%

references to openings and patterns in technical descriptions

holes and openings

No Comments

No Known Activations