Guide-yakabatana mibvunzo · Pakati Nhanho

Adapter Layers for Transfer Quiz

Tarisa kuti mabhodhoro ebhodhoro anochengetedza sei yakaomeswa base modhi, gonesa modular basa kana domain kuendesa, uye enzanisa neLoRA.

Nzira dzekutungamira dzinoenderanaAdapter Layers For Transfer
Mubvunzo 1 ye 8

Chii chinoitika kune izvo zvakafanodzidziswa huremu paunenge uchidzidzira nemaadapter?

Tarisa mubvunzo wese8 mibvunzo

Mhinduro uye tsananguro zvinoonekwa paunenge uchitora bvunzo.

  1. Chii chinoitika kune izvo zvakafanodzidziswa huremu paunenge uchidzidzira nemaadapter?
    • Iwo akaomeswa nechando nepo chete maparamita eadapter achidzidziswa
    • Vanodzimwa
    • Vanodzokororwa zvisina tsarukano
    • Iwo akapetwa kaviri muhukuru
  2. Chii chinonzi chimiro chemukati chebhodhoro adapter?
    • Imwe hombe dense layer
    • A down-projection, nonlinearity, then up-projection, ine yasara yekubatanidza
    • Musoro wekutarisisa chete
    • A convolutional image filter
  3. Zvingangove chikamu chipi chezvikamu izvo maadapter anowanzo kudzidzisa?
    • 100%
    • 50% chaiyo
    • Kazhinji pasi pe5%
    • Hapana zvachose
  4. Sei ma adapter ari nyore kubatira mabasa mazhinji?
    • Vanoda yakaparadzana yakazara modhi kopi pabasa
    • Havagoni kubatanidzwa
    • Vanodzima base modhi mushure mebasa rega rega
    • Iwe unochengeta imwe base modhi uye uchichinjanisa madiki ega-basa adapta
  5. Ndeapi mhuri yehunyanzvi iyo adapta ndeye?
    • Parameter-inoshanda zvakanaka-tuning (PEFT)
    • Data kuwedzera
    • Image compression
    • Tokenization zvirongwa
  6. Ndeupi mukana unopihwa neLoRA pamusoro peiyo yakasarudzika yakaiswa adapta panguva yekufungidzira?
    • Haikwanisi kudzidziswa
    • Huremu hwayo hwakadzidzwa hunogona kubatanidzwa mune base matrices, kudzivirira yekuwedzera per-layer latency
    • Zvinoda kudzokorora modhi yose
    • Iyo inobvisa yakasara yekubatanidza zvachose
  7. What does a residual connection contribute to a classic bottleneck adapter?
    • A second full pretrained model copy for each task
    • A mandatory external retrieval database
    • A path that adds the adapter’s transformed representation back to the original hidden state
    • A guarantee that no added computation occurs
  8. What is one benefit of stacking a domain adapter with a task adapter?
    • Merge all user data into the frozen base weights automatically
    • Avoid using any adapter modules
    • Convert a language model into a vision encoder
    • Reuse a domain specialization together with a separate task specialization