INDEX

Explanations

references to previously described methods or procedures

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

previously described

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

findpost

-0.63

 GenerationType

-0.63

RegressionTest

-0.52

 parseFrom

-0.48

optionalTypeArgs

-0.47

 ProtoMessage

-0.47

 فريبيس

-0.46

 onOptions

-0.45

 Betracht

-0.44

httphttps

-0.43

POSITIVE LOGITS

elemField

0.50

 previously

0.42

 routine

0.41

fahrens

0.41

fq

0.40

adam

0.39

 resourceCulture

0.39

ROUTINE

0.38

previously

0.38

 Routine

0.37

Activations Density 0.247%

references to previously described methods or procedures

previously described

No Comments

No Known Activations