INDEX

Explanations

components and interactions related to complex systems or structures

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

design and capability

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

formal, technical definition-style descriptions that explain what a thing is and what it does, often using explanatory clauses to state properties, purpose, or function.

oai_token-act-pair · gpt-5 Triggered by @liuliujiujiu05

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

Amazon

-0.43

 Amazon

-0.41

 Alex

-0.41

BOR

-0.40

動

-0.40

Des

-0.40

dian

-0.39

CD

-0.39

相

-0.39

POSITIVE LOGITS

">//

0.63

 CURIAM

0.58

parsedMessage

0.57

 AssemblyTitle

0.56

Portail

0.52

webElement

0.52

 cherchés

0.51

onViewCreated

0.50

ActionCreators

0.48

Derbyniad

0.47

Activations Density 0.037%

components and interactions related to complex systems or structures

design and capability

formal, technical definition-style descriptions that explain what a thing is and what it does, often using explanatory clauses to state properties, purpose, or function.

No Comments

No Known Activations