INDEX

Explanations

punctuation marks and symbols that indicate significant breaks or transitions in thought

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

continuation of quoted text

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

 kleding

-0.49

 przer

-0.48

 móvel

-0.47

 épar

-0.46

 adulta

-0.45

ไข

-0.44

 isolado

-0.44

 isolada

-0.43

 muebles

-0.43

 femininas

-0.42

POSITIVE LOGITS

TagMode

0.55

httphttps

0.51

 <<<<<<<<<<<<<<

0.48

 numerus

0.45

 HttpHeaders

0.42

UnusedPrivate

0.41

INCREF

0.41

#%

0.41

<blockquote>

0.40

Бахар

0.39

Activations Density 0.001%

punctuation marks and symbols that indicate significant breaks or transitions in thought

continuation of quoted text

No Comments

No Known Activations