INDEX

Explanations

references to additional factual information or evidence

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

additional/extra nouns

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

WindowConstants

-0.49

στη

-0.39

 totally

-0.39

apis

-0.37

 боя

-0.37

initView

-0.37

 fully

-0.35

➩

-0.35

kel

-0.35

 Reyes

-0.35

POSITIVE LOGITS

 extra

0.90

 additional

0.87

 EXTRA

0.85

 zusätzlichen

0.82

additional

0.82

Additional

0.82

extra

0.81

 Additional

0.81

Extra

0.80

 Extra

0.80

Activations Density 0.069%

references to additional factual information or evidence

additional/extra nouns

No Comments

No Known Activations