INDEX

Explanations

technical and procedural language related to troubleshooting and repairs

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

restored or improved

np_acts-logits-general · gemini-2.5-flash-lite

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

IFrame

Link

Not in Any Lists

No Comments

Negative Logits

 CreateTagHelper

-0.59

ImageContext

-0.58

tagHelperRunner

-0.55

sillon

-0.53

 bicolor

-0.52

StoryboardSegue

-0.52

 Majefty

-0.51

telles

-0.51

 localObject

-0.50

 Dominus

-0.49

POSITIVE LOGITS

 restored

0.38

 gyhoeddwyd

0.36

 verbess

0.34

Nationalité

0.32

 perbaikan

0.32

 ändert

0.32

 improves

0.32

 améliorer

0.31

 wpły

0.31

 memperbaiki

0.31

Activations Density 0.028%

technical and procedural language related to troubleshooting and repairs

restored or improved

No Comments

No Known Activations