Guide-yakabatana mibvunzo · Zvakaoma Nhanho

RWKV Linear Attention Quiz

Ongorora gedhi rekugamuchira reRWKV, kuora nguva, kiyi / kukosha kudzokororwa, inogara-saizi mamiriro, uye Transformer-yakafanana fomu rekudzidzisa.

Nzira dzekutungamira dzinoenderanaRwkv Linear Attention
Mubvunzo 1 ye 8

Chii chinonzi headline property yeRWKV?

Tarisa mubvunzo wese8 mibvunzo

Mhinduro uye tsananguro zvinoonekwa paunenge uchitora bvunzo.

  1. Chii chinonzi headline property yeRWKV?
    • Inodzidzisa seTransformer asi inomhanya inference seRNN
    • Inogona kudzidziswa chete pamifananidzo
    • Inoda quadratic key-value cache
    • Haikwanise kufananidzwa zvachose
  2. Mabhii ari muRWKV anomirirei?
    • Verenga, Nyora, Chengeta, Simbisa
    • Kugamuchira, Huremu, Kiyi, Kukosha
    • Recurrent, Wide, Kernel, Vector
    • Random, Weighted, Knowledge, Variation
  3. RWKV inochengeta sei ndangariro nguva dzose panguva yechizvarwa?
    • Nekudzima maremu emuenzaniso
    • Nekuchengeta tokeni yese padisiki
    • Nekutakura diki-yakagadziriswa-saizi yakavanzika nyika kumberi panzvimbo yekukura cache
    • Nekudzikamisa zvinobuda kune gumi tokens
  4. Ibasa rei rinoitwa nehuremu hwekuora nguva (W)?
    • Inoisa mwero wekudzidza
    • Inoverenga nhamba yezviratidzo
    • Inotora mutauro wekubuda
    • Inodzora kuti ruzivo rwekare runokurumidza kurasikirwa sei
  5. Zvichienzaniswa nekutarisa kwesoftmax, chii chinodzivirirwa neRWKV's linear-attention?
    • Iyo quadratic ese-pairi chiratidzo chekuenzanisa
    • Chero kushandiswa kwekiyi uye zvakakosha
    • Kudikanwa kwekumisikidza
    • Kushandiswa kweneural network
  6. Nei RWKV ichinyanya kukodzera kutenderera chizvarwa?
    • Nekuti inofanirwa kuona yese kupinza isati yabuda
    • Nekuti yega yega chiratidzo chitsva chinotsamira chete pane yakapfuura yakagadziriswa-saizi mamiriro
    • Nekuti haikwanise kugadzira zvinyorwa
    • Nekuti inoda GPU cluster yechiratidzo chese
  7. In the guide’s RWKV description, what does the receptance gate control?
    • How many image patches are generated
    • The size of the model vocabulary
    • How much of the accumulated state is read out
    • The amount of GPU memory available
  8. What training property does RWKV retain from Transformers?
    • Full quadratic key-value caching at inference
    • A separate generator and discriminator
    • A fixed top-k token router in every layer
    • Parallelizable training across sequence positions