INDEX

Explanations

references to independent reviews and reports regarding audits and regulatory practices

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

official reports and investigations

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

 wireType

-0.41

Искәрмәләр

-0.40

avyzd

-0.37

xnn

-0.34

MockBean

-0.34

 reciprocal

-0.33

Билгалдахарш

-0.33

 smooth

-0.32

pregunt

-0.32

 WEEKLY

-0.32

POSITIVE LOGITS

 report

0.93

报告

0.83

 Report

0.82

report

0.75

 recommendations

0.73

報告

0.71

Report

0.70

 conclusions

0.69

 findings

0.67

 relatório

0.66

Activations Density 0.186%

references to independent reviews and reports regarding audits and regulatory practices

official reports and investigations

No Comments

No Known Activations