INDEX

Explanations

references to websites

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

websites

np_acts-logits-general · gemini-2.0-flash

mentioning websites and URLs

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-16k

Configuration

google/gemma-scope-9b-it-res/layer_20/width_16k/average_l0_91

Prompts (Dashboard)

24,576 prompts, 128 tokens each

Dataset (Dashboard)

monology/pile-uncopyrighted

Features

16,384

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

 push

-0.52

 Tough

-0.48

push

-0.47

Tol

-0.47

Gil

-0.46

 Lone

-0.46

 Miel

-0.43

onAttach

-0.43

trace

-0.43

pred

-0.43

POSITIVE LOGITS

Museum

0.65

getDocument

0.60

iastes

0.59

 gebouw

0.56

 Museum

0.56

 Stadium

0.54

museum

0.53

anthemum

0.52

listdir

0.52

TokenNameLPAREN

0.52

Activations Density 0.068%

references to websites

websites

mentioning websites and URLs

No Comments

No Known Activations