INDEX

Explanations

negations or expressions of uncertainty

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

may not / might not

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

 yapılan

-0.44

 haremos

-0.40

的一种

-0.38

 идёт

-0.36

 crece

-0.36

的是

-0.36

 rimane

-0.36

 topik

-0.35

 occurs

-0.35

 queda

-0.35

POSITIVE LOGITS

 فريبيس

0.71

 disambiguazione

0.64

ThroughAttribute

0.63

 pylint

0.62

otoro

0.60

tamol

0.59

########.

0.58

testens

0.58

}{*}{

0.57

不一定

0.57

Activations Density 0.011%

negations or expressions of uncertainty

may not / might not

No Comments

No Known Activations