INDEX

Explanations

phrases and words related to preliminary results and analyses in studies or research

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

upné

-0.34

 both

-0.33

PathVariable

-0.32

Referanser

-0.32

 constantemente

-0.32

 princesse

-0.31

 actually

-0.30

actually

-0.30

due

-0.29

 each

-0.29

POSITIVE LOGITS

liminary

0.77

 Initial

0.71

Preliminary

0.71

 preliminary

0.69

 initial

0.69

初步

0.69

NameInMap

0.68

 prelimin

0.68

Initial

0.67

 المعيارى

0.66

Activations Density 0.046%

phrases and words related to preliminary results and analyses in studies or research

No Comments

No Known Activations