INDEX

Explanations

references to conspiracies and related legal terms

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

spoken communication

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GEMMA-2-9B-IT @ 20-gemmascope-res-131k

Configuration

google/gemma-scope-9b-it-res/layer_20/width_131k/average_l0_81

Prompts (Dashboard)

24,576 prompts, 128 tokens each

Dataset (Dashboard)

monology/pile-uncopyrighted

Features

131,072

Data Type

float32

Hook Name

blocks.20.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

VersionUID

-0.44

 snippetHide

-0.35

 working

-0.35

 disel

-0.32

Опис

-0.32

可

-0.31

 stunt

-0.30

 Recon

-0.30

 який

-0.29

POSITIVE LOGITS

 voice

0.58

Voice

0.56

 verbal

0.55

Spoken

0.55

 Spoken

0.55

voice

0.53

 Voice

0.53

 Verbal

0.52

verbal

0.52

parsedMessage

0.52

Activations Density 0.127%

references to conspiracies and related legal terms

spoken communication

No Comments

No Known Activations