INDEX

Explanations

instances of the word "improve" or its variations, indicating a focus on enhancement or progress

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

improving or enhancing something

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

RetentionPolicy

-0.56

 Palacios

-0.52

 Schatz

-0.52

 BorderRadius

-0.51

QPalette

-0.50

CardHeader

-0.50

 ballots

-0.49

 Mateo

-0.49

getJSONArray

-0.49

labelText

-0.49

POSITIVE LOGITS

improve

0.94

improving

0.93

Improve

0.89

improvement

0.88

 Improve

0.88

 Mejora

0.88

 Improvement

0.84

Improving

0.84

improved

0.83

Improvement

0.83

Activations Density 0.031%

instances of the word "improve" or its variations, indicating a focus on enhancement or progress

improving or enhancing something

No Comments

No Known Activations