INDEX

Explanations

instances of conversational prompts or greetings

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

interjections like ah

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

 queſta

-0.47

ItemBackground

-0.46

戬

-0.42

OpenHelper

-0.42

 desmotivaciones

-0.41

 stockbild

-0.41

 behulp

-0.39

 computadoras

-0.39

 capucha

-0.39

⟬

-0.39

POSITIVE LOGITS

thâu

0.50

 дописавши

0.50

WithFormat

0.43

archiviato

0.43

ябре

0.41

 دریافت‌شده

0.41

 übrigens

0.41

BTW

0.40

前

0.40

帖最后由

0.40

Activations Density 0.024%

instances of conversational prompts or greetings

interjections like ah

No Comments

No Known Activations