INDEX

Explanations

references to prior knowledge and new discoveries in research

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

to our knowledge

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

TypedDataSet

-0.45

fony

-0.43

 prí

-0.40

occasione

-0.39

 Beur

-0.39

PyExc

-0.39

principalColumn

-0.39

 actionMode

-0.38

-------

-0.38

AddHtmlAttribute

-0.38

POSITIVE LOGITS

出版年

0.52

ContentLoaded

0.51

rungsseite

0.49

 fread

0.47

ciun

0.47

 Fland

0.47

 extant

0.46

ngOnDestroy

0.46

 extinct

0.45

 numerus

0.44

Activations Density 0.029%

references to prior knowledge and new discoveries in research

to our knowledge

No Comments

No Known Activations