INDEX

Explanations

concepts related to correctness and appropriateness in decision-making or actions

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

PlotsExplanationShow Test FieldDefault Test Text

IFrame

Link

Not in Any Lists

No Comments

Negative Logits

 highly

-0.34

 commonly

-0.32

 weit

-0.31

含

-0.30

 mostly

-0.29

 greatly

-0.29

 severely

-0.28

 Table

-0.28

 relatively

-0.27

asa

-0.27

POSITIVE LOGITS

 giusta

0.99

 wrong

0.90

 richtigen

0.88

ReusableCell

0.87

IntoConstraints

0.85

wrong

0.84

Wrong

0.84

 Wrong

0.84

WRONG

0.84

 RIGHT

0.83

Activations Density 0.049%

concepts related to correctness and appropriateness in decision-making or actions

No Comments

No Known Activations