INDEX

Explanations

terminology related to pipes and pipelines

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GEMMA-2-2B @ 0-gemmascope-mlp-65k

Configuration

google/gemma-scope-2b-pt-mlp/layer_0/width_65k/average_l0_72

Prompts (Dashboard)

24,576 prompts, 128 tokens each

Dataset (Dashboard)

monology/pile-uncopyrighted

Features

65,536

Data Type

float32

Hook Name

blocks.0.hook_mlp_out

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

 rootReducer

-0.61

новниш

-0.59

 sözler

-0.59

 purpoſe

-0.59

 enfans

-0.57

 religieuses

-0.56

 käyttää

-0.55

λλην

-0.54

 للمعارف

-0.54

zeiro

-0.53

POSITIVE LOGITS

ThroughAttribute

0.76

 InputDecoration

0.66

findpost

0.59

phite

0.59

 calendriers

0.58

XmlEnum

0.57

 Pute

0.57

DBNull

0.56

fillType

0.55

her

0.54

Activations Density 0.045%

terminology related to pipes and pipelines

No Comments

No Known Activations