INDEX

Explanations

phrases indicating historical context or previous states of being

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

Past tense or "used to be"

np_acts-logits-general · gemini-2.0-flash

used to be / once lived

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

 diesmal

-0.84

SourceChecksum

-0.76

 ostavi

-0.75

finally

-0.74

 henceforth

-0.74

 finally

-0.73

 uiteindelijk

-0.73

depuis

-0.73

 subsequently

-0.71

 subsequent

-0.71

POSITIVE LOGITS

was

0.66

Was

0.57

 were

0.56

 wasn

0.56

was

0.54

 weren

0.54

était

0.54

Before

0.54

 Before

0.51

 étions

0.51

Activations Density 0.413%

phrases indicating historical context or previous states of being

Past tense or "used to be"

used to be / once lived

No Comments

No Known Activations