INDEX

Explanations

references to suicide and suicidal behavior

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

New Auto-Interp

Configuration

google/gemma-scope-9b-pt-mlp/layer_6/width_131k/average_l0_72

Prompts (Dashboard)

24,576 prompts, 128 tokens each

Dataset (Dashboard)

monology/pile-uncopyrighted

Features

131,072

Data Type

float32

Hook Name

blocks.6.hook_mlp_out

Hook Layer

Architecture

jumprelu

Context Size

1,024

Dataset

monology/pile-uncopyrighted

Activation Function

relu

Embeds

PlotsExplanationShow Test FieldDefault Test Text

IFrame

Link

Not in Any Lists

No Comments

Negative Logits

 suicide

-1.95

suicide

-1.74

 Suicide

-1.73

Suicide

-1.55

 suicides

-1.47

 suicidio

-1.20

 suicidal

-1.16

自殺

-1.12

自杀

-1.08

 suic

-1.03

POSITIVE LOGITS

难

0.34

 Failing

0.34

Failing

0.34

sore

0.33

hot

0.32

 jardin

0.32

zah

0.31

table

0.31

pio

0.31

hot

0.31

Activations Density 0.002%

references to suicide and suicidal behavior

No Comments

No Known Activations