INDEX

Explanations

positive descriptions and metrics of success or improvement

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

despite or although followed by a mitigating factor

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

 poveznice

-0.51

 soñar

-0.50

 للاسماء

-0.49

jména

-0.48

 borboleta

-0.47

ьаж

-0.47

تقاوى

-0.45

 académica

-0.43

 almendras

-0.43

 borbo

-0.43

POSITIVE LOGITS

verifyException

0.45

CompleteListener

0.44

postIndex

0.42

 CURIAM

0.41

 оригіналу

0.38

 auri

0.36

 pied

0.36

 rebus

0.36

 inferno

0.35

ABCDEF

0.35

Activations Density 0.109%

positive descriptions and metrics of success or improvement

despite or although followed by a mitigating factor

No Comments

No Known Activations