INDEX

Explanations

phrases related to predicting outcomes or consequences

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

that precede verbs of assistance or causation

np_acts-logits-general · gemini-2.0-flash

enable or cause positive outcomes

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GEMMA-2-2B @ 18-gemmascope-res-16k

Configuration

google/gemma-scope-2b-pt-res/layer_18/width_16k/average_l0_74

Prompts (Dashboard)

36,864 prompts, 128 tokens each

Dataset (Dashboard)

monology/pile-uncopyrighted

Features

16,384

Data Type

float32

Hook Name

blocks.18.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

 enfans

-0.58

 conoce

-0.53

 conoció

-0.52

 oublié

-0.50

 soldati

-0.49

 mourir

-0.48

 perdiendo

-0.48

 ragione

-0.48

 välja

-0.48

 gloire

-0.47

POSITIVE LOGITS

 help

1.06

 enable

0.86

 increase

0.83

 lead

0.79

 helps

0.79

 greatly

0.76

aid

0.76

 allow

0.73

 helped

0.73

 enhance

0.71

Activations Density 0.951%

phrases related to predicting outcomes or consequences

that precede verbs of assistance or causation

enable or cause positive outcomes

No Comments

No Known Activations