INDEX

Explanations

code-related syntax and assertions in testing frameworks

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

test followed by method names

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

 unſ

-0.48

 ſur

-0.47

 ſou

-0.44

 leſs

-0.43

Kaynakça

-0.42

windowFixed

-0.42

 ſta

-0.42

 itſelf

-0.41

 Inſ

-0.40

 raiſ

-0.39

POSITIVE LOGITS

 Beware

0.54

تقاوى

0.49

 beware

0.49

the

0.49

esten

0.49

DeleteBehavior

0.49

 scaring

0.47

 toning

0.47

ornos

0.46

 Freestyle

0.46

Activations Density 0.015%

code-related syntax and assertions in testing frameworks

test followed by method names

No Comments

No Known Activations