INDEX

Explanations

proper names, especially those of individuals

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

reporting verbs or statements

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

ंदीखरीदारी

-0.80

uxxxx

-0.76

 незавершена

-0.68

IUrlHelper

-0.66

 resourceCulture

-0.64

 AssemblyProduct

-0.64

__*/

-0.64

Cyfeiriadau

-0.62

 مشين

-0.61

fromnode

-0.60

POSITIVE LOGITS

 said

0.47

 explained

0.42

 nói

0.40

 stated

0.37

 statement

0.36

 powied

0.35

 boneco

0.33

explained

0.33

 stressed

0.32

 described

0.32

Activations Density 0.010%

proper names, especially those of individuals

reporting verbs or statements

No Comments

No Known Activations