INDEX

Explanations

references to COVID-19 and mentions of data encoding formats

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

"COVID" or "UTF" tokens

np_acts-logits-general · gemini-2.0-flash

COVID and UTF-8

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

er

-0.59

″]

-0.52

or

-0.50

use

-0.50

</tr>

-0.50

are

-0.48

 Lander

-0.48

</b>

-0.48

-0.47

POSITIVE LOGITS

COVID

0.84

 COVID

0.82

UTF

0.81

Covid

0.79

 Covid

0.79

utf

0.73

 covid

0.71

 ainfi

0.71

UTF

0.70

Royaume

0.69

Activations Density 0.098%

references to COVID-19 and mentions of data encoding formats

"COVID" or "UTF" tokens

COVID and UTF-8

No Comments

No Known Activations