INDEX

Explanations

negative sentiments or criticisms regarding personal actions and relationships

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

phrases expressing strong opinions or qualifying statements (often starting with "don't get me wrong" or similar qualifying language).

oai_token-act-pair · claude-3-5-haiku-20241022 Triggered by @neilrathi

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GEMMA-2-9B @ 20-gemmascope-res-16k

Configuration

google/gemma-scope-9b-pt-res/layer_20/width_16k/average_l0_68

Prompts (Dashboard)

24,576 prompts, 128 tokens each

Dataset (Dashboard)

monology/pile-uncopyrighted

Features

16,384

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

AccessorTable

-0.45



-0.45

Personensuche

-0.43

AndEndTag

-0.43

ItemBackground

-0.43

mability

-0.41

IsContent

-0.41

--}}

-0.41

 שוליים

-0.40

zah

-0.40

POSITIVE LOGITS

 تانيه

0.40

edition

0.39

uevo

0.39

 الحره

0.39

 vicenda

0.38

申し訳

0.36

 SqlCommand

0.36

Edition

0.36

 neutr

0.36

 EVERYONE

0.35

Activations Density 0.173%

negative sentiments or criticisms regarding personal actions and relationships

phrases expressing strong opinions or qualifying statements (often starting with "don't get me wrong" or similar qualifying language).

No Comments

No Known Activations