INDEX

Explanations

references to visual media and characters, particularly those involving strong themes or complex narratives

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GEMMA-2-9B @ 23-gemmascope-res-16k

Configuration

google/gemma-scope-9b-pt-res/layer_23/width_16k/average_l0_120

Prompts (Dashboard)

24,576 prompts, 128 tokens each

Dataset (Dashboard)

monology/pile-uncopyrighted

Features

16,384

Data Type

float32

Hook Name

blocks.23.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

 therefore

-0.39

 Therefore

-0.34

 infatti

-0.34

therefore

-0.32

 hence

-0.32

 appropriately

-0.31

 cost

-0.30

 Hence

-0.30

Therefore

-0.30

 often

-0.30

POSITIVE LOGITS

 pleaſure

0.64

 столь

0.63

 solch

0.60

こんなに

0.59

 tão

0.57

這麼

0.57

neſs

0.57

؟!

0.56

そんなに

0.55

 Majefty

0.55

Activations Density 0.558%

references to visual media and characters, particularly those involving strong themes or complex narratives

No Comments

No Known Activations