INDEX

Explanations

phrases that indicate relationships, interactions, or experiences involving inclusion and competition

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

 Wahrheit

-0.39

instead

-0.35

 nakalista

-0.35

 Amongst

-0.35

 saja

-0.34

 Among

-0.34

 Krieges

-0.34

 dependencia

-0.34

 Absicht

-0.34

 Existenz

-0.33

POSITIVE LOGITS

SharedDtor

0.61

'{@

0.59

 BoxFit

0.58

SourceChecksum

0.57

 endblock

0.54

elemField

0.54

verifyException

0.53

ThroughAttribute

0.53

 rootReducer

0.52

########.

0.52

Activations Density 0.174%

phrases that indicate relationships, interactions, or experiences involving inclusion and competition

No Comments

No Known Activations