INDEX

Explanations

phrases indicating the final outcome or conclusion of discussions

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

Follows abbreviations or question/answer markers

np_acts-logits-general · gemini-2.0-flash

foreign languages or terms

np_acts-logits-general · gemini-2.5-flash-lite

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GEMMA-2-2B @ 18-gemmascope-res-16k

Configuration

google/gemma-scope-2b-pt-res/layer_18/width_16k/average_l0_74

Prompts (Dashboard)

36,864 prompts, 128 tokens each

Dataset (Dashboard)

monology/pile-uncopyrighted

Features

16,384

Data Type

float32

Hook Name

blocks.18.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

 meanwhile

-1.21

 suddenly

-1.04

 consequently

-1.03

 accordingly

-1.02

suddenly

-1.00

 specifically

-0.96

 respectively

-0.94

 incidentally

-0.94

 correspondingly

-0.93

 obviously

-0.93

POSITIVE LOGITS

 boils

0.55

 ceea

0.47

 ganske

0.42

стям

0.40

 chodzi

0.40

っぱ

0.39

simp

0.39

بیشتر

0.38

 pretty

0.38

 bastante

0.37

Activations Density 0.501%

phrases indicating the final outcome or conclusion of discussions

Follows abbreviations or question/answer markers

foreign languages or terms

No Comments

No Known Activations