INDEX

Explanations

pronouns and specific contextual references

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

possessive or demonstrative

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

SourceChecksum

-0.43

providedIn

-0.43

StoryboardSegue

-0.37

MockBean

-0.36

fony

-0.35

 AssemblyCulture

-0.35

 barbaric

-0.34

si

-0.34

 verwe

-0.34

SI

-0.34

POSITIVE LOGITS

 surla

0.64

uxxxx

0.63

 مشين

0.60

 townhouse

0.57

İstinadlar

0.50

 виправивши

0.49

Tembelea

0.49

 houſe

0.47

AndEndTag

0.46

 Grundstück

0.46

Activations Density 0.002%

pronouns and specific contextual references

possessive or demonstrative

No Comments

No Known Activations