INDEX

Explanations

terms related to legal or procedural terminology

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GEMMA-2-2B @ 5-gemmascope-res-16k

Configuration

google/gemma-scope-2b-pt-res/layer_5/width_16k/average_l0_68

Prompts (Dashboard)

36,864 prompts, 128 tokens each

Dataset (Dashboard)

monology/pile-uncopyrighted

Features

16,384

Data Type

float32

Hook Name

blocks.5.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

 cauſe

-0.80

 quæ

-0.79

 pleaſure

-0.78

 ſon

-0.74

 Diſ

-0.73

 reaſon

-0.73

 purpoſe

-0.73

 Jefus

-0.72

 houſe

-0.72

 perſon

-0.72

POSITIVE LOGITS

 vitro

0.79

hoc

0.77

 Vitro

0.69

 <<<<<<<<<<<<<<

0.66

 vivo

0.66

 priori

0.65

 facie

0.61

 cappella

0.61

hamdulillah

0.59

TintMode

0.59

Activations Density 0.413%

terms related to legal or procedural terminology

No Comments

No Known Activations