INDEX

Explanations

instances of the word "was" and its variations emphasizing positive experiences or attributes

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GEMMA-2-9B @ 27-gemmascope-res-16k

Configuration

google/gemma-scope-9b-pt-res/layer_27/width_16k/average_l0_118

Prompts (Dashboard)

24,576 prompts, 128 tokens each

Dataset (Dashboard)

monology/pile-uncopyrighted

Features

16,384

Data Type

float32

Hook Name

blocks.27.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

 Parcelable

-0.37

guém

-0.36

epresidente

-0.36

 jefe

-0.35

 espé

-0.35

letín

-0.35

 caché

-0.34

ריים

-0.34

 perd

-0.34

 Edad

-0.33

POSITIVE LOGITS

 snowing

0.63

 raining

0.59

AddTagHelper

0.52

 esternos

0.51

AndEndTag

0.50

raining

0.49

httphttps

0.48

pain

0.48

richTextPanel

0.47

MessageTagHelper

0.46

Activations Density 0.175%

instances of the word "was" and its variations emphasizing positive experiences or attributes

No Comments

No Known Activations