INDEX

Explanations

expressions related to performance and effectiveness

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

performance of

np_acts-logits-general · gemini-2.5-flash-lite

New Auto-Interp

Configuration

google/gemma-scope-9b-pt-res/layer_24/width_131k/average_l0_97

Prompts (Dashboard)

24,576 prompts, 128 tokens each

Dataset (Dashboard)

monology/pile-uncopyrighted

Features

131,072

Data Type

float32

Hook Name

blocks.24.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

atguigu

-0.40

 kiss

-0.36

<h1>

-0.36

 knot

-0.35

 necessários

-0.35

gas

-0.35

sendFile

-0.35

 gift

-0.34

uus

-0.34

FileDialog

-0.32

POSITIVE LOGITS

 performance

1.19

performance

1.08

 performances

1.03

PERFORMANCE

1.02

 Performance

1.01

 PERFORMANCE

1.00

Performance

0.99

 desempenho

0.89

 Performances

0.87

Performances

0.83

Activations Density 0.028%

expressions related to performance and effectiveness

performance of

No Comments

No Known Activations