INDEX

Explanations

mentions of sources in documents

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

source followed by colon

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

rostis

-0.52

transQ

-0.49

copg

-0.48

 igång

-0.48

ligheter

-0.48

rboles

-0.47

vician

-0.47

adrez

-0.46

 Italijani

-0.46

 fören

-0.45

POSITIVE LOGITS

source

0.96

 source

0.96

Source

0.96

 Source

0.94

 SOURCE

0.82

SOURCE

0.75

 sources

0.71

Bronnen

0.71

sumber

0.68

ource

0.68

Activations Density 0.004%

mentions of sources in documents

source followed by colon

No Comments

No Known Activations