Guide-yakabatana mibvunzo · Pakati Nhanho

Muenzaniso Kubvisa uye Kuba Kurwisa Quiz

Nzwisisa kuti vanorwisa vanotevedzera sei mhando dzebhokisi dema kuburikidza nemibvunzo, zvinoburitswa neAPI, uye kuti vanopa vanogona kuona kana kudzivirira kutorwa.

Nzira dzekutungamira dzinoenderanaModel Extraction And Kuba Kurwiswa
Mubvunzo 1 ye 8

Ndeipi pfungwa huru kuseri kwemuenzaniso wekutora kurwisa?

Tarisa mubvunzo wese8 mibvunzo

Mhinduro uye tsananguro zvinoonekwa paunenge uchitora bvunzo.

  1. Ndeipi pfungwa huru kuseri kwemuenzaniso wekutora kurwisa?
    • Kubvunza modhi uye kudzidzisa chinotsiva kutevedzera zvabuda
    • Kubira museva kukopa uremu mafaera zvakananga
    • Kupa mushandi chiokomuhomwe kuti aburitse dataset yekudzidziswa
    • Kuwedzeredza iyo API kusvika yarovera uye yarasa ndangariro
  2. Sei kudzorera zvizere zvingangoita mavheta zvichiita kuti kudhirowa kuve nyore pane kudzosera chete top-1 label?
    • Iyo encrypts mhinduro, ichinonotsa anorwisa pasi
    • Inoburitsa ruzivo rwakawanda pamubvunzo wega wega, saka mibvunzo mishoma inodiwa
    • Inovanza muganhu wesarudzo zvachose
    • Iyo inowedzera server mutengo saka vanorwisa vanorega
  3. Ndeipi nzira yekudzivirira inoderedza zvakananga ruzivo rwakaburitswa pamubvunzo wega wega?
    • Kuwedzera mamwe maGPU kune server
    • Kuwedzera iyo muenzaniso parameter kuverenga
    • Kudzorera zvakatenderedzwa kana pamusoro-1-chete zvabuda
    • Kuburitsa data rekudzidziswa pachena
  4. What did the 2024 production-language-model extraction paper report about GPT-3.5 Turbo?
    • It reconstructs every GPT-3.5 parameter exactly for under $20
    • It obtains the model’s training dataset directly
    • It extracts a complete model checkpoint from the API
    • It recovers the embedding-projection layer; the paper estimates under $2,000 in queries for GPT-3.5 Turbo
  5. Sei imwe yekutsiva yakabiwa ine ngozi kupfuura kungorasikirwa nemari?
    • Inogonesa kurwisa chena-bhokisi kurwisa sekugadzira mienzaniso yeanopikisa
    • Inovandudza kurongeka kwemuenzaniso wepakutanga
    • Iyo inodzima iyo yekutanga modhi
    • Zvinodiwa kudzidzisazve yepakutanga
  6. In ActiveThief, what does the combined DFAL + K-center query strategy aim to balance?
    • Boundary-proximate informative samples and reduced redundancy among selected samples
    • Exact recovery of all secret model weights and a guaranteed substitute
    • Random sampling only and avoiding all substitute-model predictions
    • Higher API throughput and removal of the query budget
  7. What can API rate limits do against a model-extraction attempt?
    • Prove that no substitute model can be trained
    • Remove all information from each permitted prediction
    • Make large-scale querying slower or more expensive
    • Make model outputs impossible to compare
  8. What does the DynaMarks paper propose for a model-extraction defense?
    • It adds a logo to every generated answer
    • It encrypts all model weights at rest
    • It guarantees ownership of every copied production model
    • It can embed a signature in a surrogate model by dynamically changing API outputs using secret parameters