INDEX

Explanations

specific terms and concepts that relate to data, information, rules, evidence, and operational specifics in various contexts

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

-0.42

too

-0.40

 very

-0.39

 beautiful

-0.39

 less

-0.39

 bulundu

-0.38

 well

-0.37

++.

-0.36

anthene

-0.36

 just

-0.35

POSITIVE LOGITS

MessageOf

0.61

 سكانية

0.61

تقاوى

0.59

 ProtoMessage

0.58

Personensuche

0.55

...?"

0.54

 Administrativna

0.54

Хьажоргаш

0.53

Quelles

0.52

 podjela

0.52

Activations Density 0.050%

specific terms and concepts that relate to data, information, rules, evidence, and operational specifics in various contexts

No Comments

No Known Activations