INDEX

Explanations

references to feelings of something being wrong or problematic

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

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

Itr

-0.47

nologue

-0.47

smöglichkeiten

-0.46

 Fazit

-0.45

träger

-0.45

 niej

-0.42

 Jerusalén

-0.42

şört

-0.42

 käyt

-0.41

 praktijk

-0.40

POSITIVE LOGITS

 amiss

0.58

 propOrder

0.57

aarrggbb

0.54



0.53

 이상

0.52

異常

0.52

おかしい

0.50

 anormal

0.49

 stimmt

0.47

 abnorm

0.47

Activations Density 0.023%

references to feelings of something being wrong or problematic

No Comments

No Known Activations