INDEX

Explanations

words that indicate certainty or affirmations about various subjects

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GEMMA-2-9B @ 38-gemmascope-res-16k

Configuration

google/gemma-scope-9b-pt-res/layer_38/width_16k/average_l0_128

Prompts (Dashboard)

24,576 prompts, 128 tokens each

Dataset (Dashboard)

monology/pile-uncopyrighted

Features

16,384

Data Type

float32

Hook Name

blocks.38.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

 actually

-1.35

 realmente

-1.33

 essentially

-1.32

 basically

-1.32

 obviously

-1.25

actually

-1.25

 evidently

-1.22

essentially

-1.21

obviously

-1.20

 eigenlijk

-1.20

POSITIVE LOGITS

 parlant

0.44

ً

0.39

 hablando

0.37

 speaking

0.36

keits

0.36

ঃ

0.35

 destinées

0.35

حوالہ

0.34

 jakieś

0.34

Erreferentziak

0.33

Activations Density 0.806%

words that indicate certainty or affirmations about various subjects

No Comments

No Known Activations