注意力机制和Seq2seq模型.ipynb

{"cells":[{"attachments":{"a1.PNG":{"image/png":"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"},"a2.PNG":{"image/png":"iVBORw0KGgoAAAANSUhEUgAAAg4AAAGXCAYAAADbOtgvAAAAAXNSR0IArs4c6QAAAARnQU1BAACxjwv8YQUAAAAJcEhZcwAAEnQAABJ0Ad5mH3gAAEZYSURBVHhe7Z0HnB3FfccxVXSBAAGid1sCGUJvwpZjIGAgBJlmBJgmepdjGVGC6NX0jkEUQzA2nYBNM8Kml0BMU8AGgxE2GAhgOwmb/Ab9j9He3N2+dzPv5u37zufz/bx7u7O7s3P7Zr47Ozsz09/+9rcCAAAAoAqIAwAAAFQGcQAAAIDKIA4AAABQGcQBAAAAKoM4AAAAQGUQBwAAAKgM4gAAAACVQRwAAACgMogDAAAAVAZxAAAAgMogDgAAAFAZxAEAAAAqgzgAAABAZRAHAAAAqAziAAAAAJVBHAAAAKAyiAMAAABUBnEAAACAyiAOAAAAUBnEAQAAACqDOAAAAEBlEAcAAACoDOIAAN04/fTTi8GDBxfTpk0LrgfIGV23jz/+eHBdDFLvP3cQBwDohsRhpplmQhygLVl66aWLgw46KLguBpLqlPvPHcQBALqBOEA7k7pi128DcQCAtuemm25yhdkrr7zSbd3EiRPdevt+7733umVbbbVVMXbsWPe3H78sDpdddlm3OKK8X6EmXNu30lNeL8rH1/7LcaCz0LVm17CuC10f5Ws5dL1pO21jjw70t65da3Xw96u/7bqrun/Fsf3bscr79+N3AogDQE1QAaYCTZW+v1wFoZZbgajCUXdko0aNcgWfPrV+5MiRXduUxUGVuwpKW29omdbZdwmB9m13fCqgy2lSOkLHV7osDnQeulZ0XehT14WuLV0XuqYsTvl6E6rYFU/yqetV6/Vd17P+1jJdf9q3ltl1p31pmS8Pof1LGPravx+/E0AcAGqECkUVhv4yFcQqEPW3CjkrnP04KkhVGFoh2ow4KK4KU+EXpipwtS+7I9SxFcfWCxMIfxl0Dna9le/2dX35151/vRm+ONgyXeO6puy77b+8bZX9++Jgy/Td33+ngTgA1AirpE0ATBT8u3mt8yt2/W0Fq93dNSMO2rZcwArtQ8stDSYp+u6nAzoXyWRIHMvSGarYGxEH249h+7frEHGoBuIAUCNUAPqFpj2mMJGwOKq0VVDrzl/xFUf0RxxsG6F9lrFWDu1Tx7W41nRsx4LOI1Rhi3Kl3V9xsO+Gya5d94hDNRAHgJqhgk8Fp/5WZe0/lrBK2ypyFagqGE0wehMH26dPSBz0qf2VsX0b+m4Co+20f19woHOw67G83CptXT/6ruutHC9UsWt/jYiDXXeIQzUQB4CaYQWdPlWA+gWe/i4XgsIK1p7EQYVkWRy0TnGsoFXhru/athxPy6xwDkmEpcsqCOgsJI/l60uUr0tJb1kcrPL3r+mexKEsphJX/7ih/Yd+M/qOOABArdCdkwpEYZW/CBWCEgzF13KruMviYN/9il0Fp5b5d2h2XP+YVjhb4W+tF34hXq4goLMISaeuD1Xkft+H8rWj60wVvbYti4MvAHZ9+fsysfYFwATG37+1iJX3X26Z6CQQB4AaYgVl+e5JBaFJgtapYNZ3Ve5aZgW3bW8C4G9njzpUoAq/AFXFr3X+/vW3XzhrX1rWWxzoPHQd2TVhFbiuDbsGhbUuCIuj60affsVulb2uWUmAXc8mto3uX9/9/ds1a/u35Z0C4gBQQ1QYqqDrqVDTHZ4KRBWouvPSMsW3O35t5xeUhpZJMvSpYyi+bWPYsS1eT2nQsRVHlPcBnYmuRf+a8Ct1P46uHV2/dt2ErjMtUzztw8TBvzb72r9/Xfa1f395J4A4AABArfHFIbQeGgNxAACAWoM4xAVxAACAWqNHc+X+DNA8iAMAAABUBnEAAACAyiAOAAAAUBnEAQAAACqDOAAAAEBlEAcAAACoDOIAAAAAlUEcAAAAoDKIAwAAAFQGcQAAAIDKIA4AAABQGcQBAAAAKoM4AAAAQGUQB8ieKVOmDBih9LQzb731VvA8W8Gjjz4aTFO7EzrXVvHJJ58E0wSQEsQBsmbDDTd08+gPFIcffngwXe3Ihx9+WAxZaKHgebaKG264IZi2duWEE04InmerWHXVVYPpqgsHH3xwMXz48AHjvvvuC6ar00EcIGuGDh3qCsil1tuypSzylXXdcXfcccdgutqRl156yZ3T7PMMDp5zSuZfcmV37LPOOiuYtnblgAMOcOe10IprBM87JTPPOnsxyyyzBNNVF5S3A8khhxwSTFengzhA1pg47H3fZ8Xe9xctY4vT73HHraM4LPLldYLnnJI1dz/OHbuu4jB64nXB807JoPmGdIQ4zLfAQsXkR95pKeOOOd8dG3EIgzhA1iAO8UAc4oM4pEV5O/+QRYpbXypaymGnTXbHRhzCIA6QNYhDPBCH+CAOaVHeIg75gThA1iAO8UAc4oM4pEV5izjkB+IAWYM4xANxiA/ikBblLeKQH4gDZA3iEA/EIT6IQ1qUt4hDfiAOkDWIQzwQh/ggDmlR3iIO+YE4QNYgDvFAHOKDOKRFeYs45AfiAFmDOMQDcYgP4pAW5S3ikB+IA2QN4hAPxCE+iENalLeIQ34gDpA1iEM8EIf4IA5pUd4iDvmBOEDWIA7xQBzigzikRXmLOOQH4gBZgzjEA3GID+KQFuUt4pAfiANkDeIQD8QhPohDWpS3iEN+IA6QNYhDPBCH+CAOaVHeIg75gThA1iAO8UAc4oM4pEV5izjkB+IAWYM4xANxiA/ikBblLeKQH4gDZA3iEA/EIT6IQ1qUt4hDfiAOkDWIQzwQh/ggDmlR3iIO+YE4QNYgDvFAHOKDOKRFeYs45AfiAFmDOMQDcYgP4pAW5S3ikB+IA2QN4hAPxCE+iENalLeIQ34gDpA1iEM8EIf4IA5pUd4iDvmBOEDWIA7xQBzigzikRXmLOOQH4gBZgzjEA3GID+KQFuUt4pAfiANkDeIQD8QhPohDWpS3iEN+IA6QNYhDPBCH+CAOaVHeIg75gThA1iAO8UAc4oM4pEV5izjkB+IAWYM4xANxiA/ikBblLeKQH4gDZA3iEA/EIT6IQ1qUt4hDfiAOkDWIQzwQh/ggDmlR3iIO+YE4QNYgDvFAHOKDOKRFeYs45AfiAFmDOMQDcYgP4pAW5S3ikB+IA2QN4hAPxCE+iENalLeIQ34gDpA1iEM8EIf4IA5pUd4iDvmBOEDWIA7xQBzigzikRXmLOOQH4gBZgzjEA3GID+KQFuUt4pAfiANkDeIQD8QhPohDWpS3iEN+IA6QNYhDPBCH+CAOaVHeIg75gThA1iAO8UAc4oM4pEV5izjkB+IAWYM4xANxiA/ikBblLeKQH4gDZA3iEA/EIT6IQ1qUt4hDfiAOkDWIQzwQh/ggDmlR3iIO+YE4QNYgDvFAHOKDOKRFeYs45AfiAFmDOMQDcYgP4pAW5S3ikB+IA2QN4hAPxCE+iENalLeIQ34gDpA1iEM8EIf4IA5pUd4iDvmBOEDWIA7xQBzigzikRXmLOOQH4gBZgzjEA3GID+KQFuUt4pAfiANkDeIQD8QhPohDWpS3iEN+IA6QNYhDPBCH+CAOaVHeIg75gThA1iAO8UAc4oM4pEV5izjkB+IAWYM4xANxiA/ikBblLeKQH4gDZA3iEA/EIT6IQ1qUt4hDfiAOkDWIQzwQh/ggDmlR3iIO+YE4QNYgDvFAHOKDOKRFeYs45AfiAFmDOMQDcYgP4pAW5S3ikB+IA2QN4hAPxCE+Jg4rjN6xWGefUxrmK1uPc3kTWtcXsw6aqyPEYdBc8xS7HXlKS9loix3csRGHMIgDZA3iEA/EIT4mDgNFJ4jDQII4hEEcIGsQh3ggDmk46aSTmmLvvfd2ebLqqqsG11fh7rvvDqapLtx1113B867KzjvvHFxelVCaAHGAzEEc4oE45MXPf/5zlyf/9E//FFwP/eP5558v/vEf/zG4DvoH4gBZgzjEA3HIC8QhLUcddZTL3zfffDO4HpoHcYCsMXFYa4/j/59JDaHKauXNdw+u6wttp+PWURzmXniJ4Dn3xYhtD2w6P4etMdodG3H4AsQhLXPMMYfL36OPPjq4HpoHcYCsWWyxxdyPf6DQM9JQutqRV199NXiOreScc84Jpq0TQRzSceedd3Zdc6usskowDjQP4gBZ88gjjxRnnHFGU6y//vqu4Nh1112D66vwX//1X8F0tSs//elPg+dZhcUXX9zlp+7gQuv74oc//GEwTZ0K4pCOsWPHdomD+NWvfhWMB82BOEBt2WeffVyhccMNNwTXQ2OMHDnS5efrr78eXA+NgTik4c9//nMx++yzd0mDOPjgg4NxoTkQB6gtiENcEIe4IA5puPjii12+LrHEEu5TqK9UKC40B+IAtQVxiAviEBfEIQ1f+9rXXL6uu+667nPttdd2nzfeeGMwPjQO4gC1BXGIC+IQF8QhPv/+7//u8lStDdtss437+/vf/777ZEyHeCAOUFsQh7ggDnFBHNKg6/SII45wnaNnnnnm4r333ivmmWee4tvf/nYwPjQO4gC1BXGIC+IQF8QhLdbicP/99wfXQ/MgDlBbEIe4IA5xQRzSgjikA3GA2oI49M5f/vKX4tNPPy0+/vjj4PoyjYjDW2+95Z43P/vss8UzzzxTPPXUU8UTTzxRPP7448Wjjz7q3qvXGB0PP/xw8dBDDxUPPvigK+Dvu+8+V6Hec889xb/927+5SY7uuOOO4vbbby9uvfXW4pZbbnFjUdx8883FT37yk+Jf//Vf3f/3xz/+cXHdddcV11xzTTF58uTiqquuKn70ox8VV1xxRXHZZZcVl156qettf+GFFxYXXHBBcd555xXnnnuuG1tCo1meeeaZxemnn16ceuqpxSmnnOImODrhhBOKSZMmFf/yL/9SHHfcccUxxxxTTJw40fGDH/zAPTv/53/+52L8+PHFkUceWRx++OHFoYce6mZUPPvss4P54tOIOCg/vvWtbxVbbrllscUWWxSbb755sdlmmxXf/OY3i7//+78vRo8e7ToFbrLJJsXGG29cbLTRRsUGG2xQrLfeeq6ToDoIrrnmmsXf/d3fFauvvnrx1a9+tVhttdWKESNGFMOHDy++/OUvu4GSVlpppWKFFVYoll9++WLZZZctll566WKppZZyfQaGDRvmBmRbdNFF3VsKCy+8cLHQQgsVCy64YDF48OBi/vnnL+add95i7rnnLuaaa65ikUUWccdTBa68fO2114LnlgrEIR2IA9SWuoqDKmRVcKpAVAmoIB8yZEixwAILuAJ8vvnmc890VYDPOeecxaBBg9x77bPNNpubhvlLX/qSyxcfVaqhY/k0Ig46VvkYnUYoX3waEYe99tqr2/4HGl1H6kMw66yzuutLQzzrepM06PoLbbPVVlsVv/zlL4PnGBvEIR2IA9SWuonD7373OzcEtl8QGxICVdYqwCUKKrwlDroDlEhIKCQWEgzdJepuUMKhO0iNCKm7+NAxfRp9VLHhhhu6u99Ro0a5u2HdFX/jG99wd8m6W9Zds+RHd9G6m956661dz/dtt9222G677YoxY8YU22+/vZsvZKeddiq+853vuBEBNRLo7rvvXuyxxx7Fnnvu6aanHjduXLHvvvsW+++/f3HggQcWBx10kLvzVwuAWgLUIqCWAbUQTJgwwbUYqOVAo2CqJUEtCscff7xrYTjxxBOLk08+2bU8nHbaaa4lQi0SakVQC4VaKs4//3zXcnHRRRcVl1xyiWvRuPzyy4srr7zStXSE8qNMo48q1DKjSnfKlCmutebXv/518dhjj7lWnCeffLJ4+umnXevOc8895+TyhRdeKH7zm98UL774YvHyyy8Xr7zySjF16lR356//4W9/+9vijTfecJNAqYXo7bffLt55553i3XffLf74xz+6ToXvv/9+8cEHHxQfffSRG0X1k08+ca1Uf/3rX4Np9FGLlo75i1/8wv1f1DJh16vyOLRNTBCHdCAOUFvqJA6qoFT563yWXHJJVyHefffdriJQYR7aJjb0cYhLJ/Zx0GMfnbOQHIbixAJxSAfiALWlLuKgO3IrbHVnrbvAULzUIA5x6URxEA888IDrU6FzV8tQKE4MEId0IA5QW+ogDocddpg7B3Viu+2224JxWgXiEJdOFQehRyvWgqbHQqE4/QVxSAfiALWl3cVBbwoo/eqv8PzzzwfjtBLEIS6dLA5Cb8no/IXeuAnF6Q+IQzoQB6gt7SwO6sCm192Ufr1SGIrTahCHuHS6OAh1VFUefP3rXw+u7w+IQzoQB6gt7SwOemNAaVefhtD6gQBxiAvi8DkaX0L5oLEzQuubBXFIB+IAtaVdxUGvu+k1SqX997//fTDOQIA4xAVx+BxV7MoHEfORBeKQDsQBaku7ioNGQ1S6c5vND3GIC+LwBSkeWSAO6UAcoLa0qzjYu+6xm277C+IQF8RhRmI/skAc0oE4QG1pV3HQSIlKd5XRHFsJ4hAXxGFG/EcWel0zFKcREId0IA5QW9pVHDQss9Kt4YRD6weKuouD+pNoqOYYaHjm0DF8GhWH0HFaxX/+538G0xSbmI8sEId0IA5QO2wc/XYVB80oqHRrnP/Q+lZjHTTrJg6a90EVlWaKjD0pl4YFDx3TpywOH374Ybc4ht6uKR+j1WgOFM2kqYmqNAZDKJ0xiPXIAnFIB+IAtUPT+6rQ0HTDmsFPkxBpgh1NGRyKnxvLLbecK/A02VBofSux5mNNAKUJsTQboqa71rLddtstuE3uqBn829/+tjsHQxOAaRhkTUcdA02+FTq2MDHUlOG6PvX9u9/9rquY//SnPwW30WOr0HFahaTR3vQxlG7NoRJKb3/Q1Op2jP48skAc0oE4QK3Qq4yaDVIFxi677OI+Dz74YPepmRhD2+SGJEezXYbWtZqHH37Y5Z2GvLYWB1WK+tSskqFtckYVndIuJELHHntsyx8JaYZPHV/zNOhTb89odFD9rZkqQ9vkgmbUlIivtdZaXfmo2U9jT5Ud45EF4pAOxAFqx1577eUKjPXWW899WtPn5MmTg/FzQ3ehmvo6tG4gWH311V3+rbDCCu5TU3LrU1M2h+LniqbUVrqFptP++OOPg/FSc8cdd7g0SMb0adenphMPxc+V22+/fQaB0PTlEvdQ3GYoP7LQVOL33HNPt3g9gTikA3GA2vGLX/zCFRg2ZLNQU3Qobm5MmzbNpVeVdGj9QHDKKae4NElmLD/1GCgUN1cuvPBCl271ZbjuuuuCcVrJ8ssv35WXQ4cOdZ8avyMUN3dOOOEEJ7s6Bz1mu/7664PxGqX8yGLxxRd3f7/11lvB+GUQh3QgDlBLvvKVr3QVOiKnoZt74+WXX3bp1Z1caP1AoOZppUmPTyw/L7/88mDcHHnqqae60n3ttdcG47SaH/zgB11pEnpsEorXLvzHf/yHe+Ri56M+JDE699oji9VWW811ItXfN998czBuGcQhHYgD1BIbRMl44IEHgvFy49FHH3Xp/eY3vxlcP1CoJ73l5ZxzzlnpdcNcGDNmjEv3xIkTg+sHAr3iaPkpDjnkkGC8duPqq692b5TonGafffbi1FNPDcargvrQaIZYe6SjPkr6PPLII4PxyyAO6UAcoJbobkeFhlhqqaWCcXLk7rvvdmnWHVto/UCh5mfLz+233z4YJ0esc+ewYcOKTz75JBhnoNBroJanv/71r4Nx2hG9VmodP8X666/vHjuE4vaGbb/ddtt1/S023HDDYPwyiEM6EAeoLfbam14lDK3PEaugNQZFaP1AYj3/9TpmaH2OWIdIvT0RWj+Q2Bse6hcQWt/uSBYkDTpHIZn44IMPgnFDXHrppa51S9ta/wY9LhNVJBBxSAfiALXlRz/6kSs4WjXqXQwuuOACl2ZVeKH1A8kBBxxQrLzyysF1uWLN3DGGMI7N+++/75rzzzjjjOD6uqDHFTpP/R/0GOOqq64Kxgvx/PPPF6NHj3bb+qgDdCi+D+KQDsQBaoteDWuXsRsM9VBXYXfyyScH1w8kavZvt7EbZp111qzfqNFYIzlNnZ4KPTr0B91SR8oXXnhhhjh33nlnj+NYHHXUUV3bij322CMYzwdxSAfiALXi1VdfdXc0eoXw0EMPLb72ta+5u/ezzjrLDT393nvvBbfLBXX8UmF38cUXB9e3mieeeKK47LLL3OMetTiof4M6GZ5//vmuoA9tkwv2NohaHULrBwJ1Kv3JT35S/PCHP3RvDGy22WauY6TGKrjyyivd2wmh7eqCHsXZq6h6hXPSpEluuYRByzRKZXkbQ+NG2OiVoRsCjbqpV231SESSYq9jr7vuusXuu+/uxu7QWBDl7aBxEAdoe1RBqBC2Pg19oTcWcm0etlEZB/Kd/l/96lfFfvvtN8NYAz2hUTr11kIurzn6PPPMMy6Nm2yySXB9KznvvPNcZacWkHIelll11VWdQL744ovBfbU7agnUYFF2vhroSf8rSYO+77///sHthORg5513LoYMGdK17KabbnLjitj++mKJJZZwx2CW1+ZBHKBt+eMf/1h8//vfn2F8gWWXXdZVerqDUx+HW265xd29H3fcce4uxIajFuqUpoGBQvseKOxd9XvvvTe4PiV6RdCGQzY0aqQqMYmW5ECFtCpBNR1vuummM8RVb/ef/exnwX0PBOrXoHTFmGmxWXQNamIoyyN19lMzvR756NrTHBTWQqa8Vh8Siys0XPobb7wR3He7o2GqJXV2rjvssEPX77OvFje1eAm//4PmqNE+9t57b/f7Uf8I3VToFWddl+ogu84663TF17wr3/ve97JvhcwRxAHaEg3bq0FzrBBQM7oqiipBw9Z+61vf6tpWFUsudx9WEOoRQWh9Ks4999yu/Bg8eHBx4oknFlOnTp2eYz0HvXqnR0J+a4+G/A4do9VoDgqlZyBGuZTU+teYpkpXU/v//M//TM+5noMEzh5Zibnnnru48cYbg8epA3pEM88887hz1bWnTz3GUMtXKL7wx2lRC42Gk68aHn/88RleF5XYSS5Cx4EwiAO0Heecc07Xj14zNGrOhGaCxkzYYIMN3H401kMj4+CnQjM0Kj26UwqtT4GesVt+akRDvTLXTLjiiiuKRRdd1O0nBxlTxaO0qB9BaH0qNCaDjVyqsRo00mEzQfmn1jP736jjbOh47Yz6HyyzzDLF5ptvPsO4FkKtL6HXLnfdddeuOGq1aTa89NJLXS18mplUrT7lY0EYxAHaCr1tYIXGaaedNr0I6F+QfNg+JROh47aCv/71r675tJVvAdijCTURq9m8v0GvvuqRhfapwrjqvAIpeOihh1w6Wvlmje5mbbwLDVykibT6G2yeDbHlllsGj9uurLLKKl3nFuIf/uEfZoivRxFaLtHX2xIxgj/5mX4D/vEgDOIAbYOer9sPXE3rMYN1rtJ75q+99lrw+KmxUSOrjozXX9TvQ8dbeumlXfN4zLDjjju6fetOMnTsVqAe9kpDqwbT0rNyGzdCd9Ixg57Z20RS7TRPSBX0Oqoe4+jxg/p/qAVC52noBkHx7JVMTQCnfh8xg7ViStqfe+65bmmEGUEcoC149tlnuzpOqVNUiqBXtrR/vcIZSkNqbDrwM888M7g+JuroqGMJzeMRO/z3f/9315TL6n8SSkNqrI+BKqTQ+thsvfXW7niq/FKEa665put/9uCDDwbTUBf0euZtt93WJRM22ZXQK5UpgvV70FseoTTBFyAO0BbstNNO7ketjngpg70SJjkJpSMV6l+h4+qushX9G6wvhcZoSBU0JoHu4HScgXh/3nrst+Lxkw0Vro56H3300fQciB/0ZoCOU27CrzvW/0GTaKUM1vKoFs1QOuBzEAfIHr22pR/zQgstFOWZcW/BKnA1l4bSkgpVODpuK1obrFlW41mkDtYnRXeNobSkwiY5Gzp0aHB9bNZYYw13PA0yljqstNJK7li33nprMC11Q69m6nxHjRo1PQfShZ///OfuWO00Md5AgDhA9tiUzrE6Q/YVrMnZnq2mRu/q63it6vhm0x5rvP9WBDteMzMkNos9dlILVWh9TNTnQMfaeOONp59x2qBWIjteKD11w8a20EilrQiSXB1Po3uG0gOIA2SOxgnQAE/qqf7ZZ59N/2mnDapQVXBoTIVQmmLx9ttvu1EXdSy9q//yyy8H48XEzk3N+K0KGoxLx2zVxF3qGa/jaUCgd955JxgnJjbts8ZaaFXQ6Ic6Zt1HP3z66afdeapFrtGgTs7avtGgMWJ0zHab56aVIA6QNfYmxbbbbjv9Z92aYM/mp02bFkxXf1HlZoW/mp5bNVW1RsrTMfUIoVXBCn89pw6lKSYaBMw60bZqWHE7Xsq+DeVgHWlzmdMkFfaoa/z48dPPvHrQ/DTattGgGxSN8KkbFt24hNLV6SAOkDU2d4Pm5q8S9G63KigVGn7QTHm6y9ZUxlWCdcaMNQeDZgLU64Hf/e533euP2rfQYDYacCm0TQpGjhzpjqu5AaoE5aPy0x/jQXmovFSeVg16hU7H1SRkoXTFQGlccMEF3XH0+CcUJzb2Cq2G364Sesq7UD73FjSolI4roQ6lK2esRUhp72tETBvWXI+5Gg3NioOCPa5Q60MoXZ0O4gBZo97j+gFrNL6qwabTtWZKK0DKMtFbsOmte+vnoPf2NWqlZOXHP/5xcfbZZ7u5M/R8XelWh7nFF198hrk0DL3yddFFFwX3mxJNDqRBphoJqtA0FLBJl0aa1PdGBuCxVyM1KFMoXf1Bg06NGzeuK297myQpNhouWcc87LDDpp9p30H5pm2UjwpqUld+NiJiEjDtY+211w6mK2ckWzZIllAHVr0KOWXKlG5xTXT1emajoT/iYJNwaTTUcpoAcYDMsd7qekWxalAFp7cihBXKGh2ykWAd3nTHo3fJVRlpeFoNUa1ZI21s/SrosYdGyNMdlgqiVg4n7aOmdKVn2LBh08+yWlAeajtVbHa32IiEKWjiIW0nsdLbABrwR53d9IhGgxup74UqVInFww8/7F7flCxqJEY9fpAEaiwPTVykeTx0J6hX5vQc2gZGmn322V26QueeCuu/ceqpp04/02rBl1v97YtZlaDpubX9IossUmiabqHHeppVVXfxQm94SGiFXhdVi5dQK5rGhND8DkKvOGqiLaFJuSRDQteqfgdCHTLV6nfJJZc49IhE4is0suUFF1zg0GvMmgRN/xuhN3jUyVBIrPX/ERreWaNAlif10vlonf1GJBVaXmWODwXlpf32hbb1v1dt0dENg7at4zDfMUAcIGtsIisNx9xIsLs6FcgqMBoplBVUqWn7nphjjjncK1u649PdtJ45T5w40RWaKsRV+emuMKdnpHaXqgmpGg1299bonbEFTYXs519M9Dxad6y/+93vguedEnsjptHxBUxulZ/avmqF5ge1HPn5UDf0WE9SuPDCC08/476DhEctOUItZdqPfRdVW8msBUvbhP7vnQ7iAFljUxL/4Q9/mP6Trh7srq7Ru2MFG1lRLR66+9CdmcZ40F2v0hJKa+5oxkadk6YTbyZYJddMT3UriPUK4S677OJEQsNSa6pzvZWg1hj9vyRhakXQox619mhAHr3dotE8ta2G41ZnUm2rQl3TJf/lL38Jnm8r0PTYOi/dUTcaTMZUwTUadAeubSVNe+yxh0P9gSSwQi08Gmpb+S723XdfN2GWUOuZRvMUBx54oJMuCZBQnuoVVqHHL2qyF0cccYSbsVOdFIU62eotGaFWJI3sKDRJmoaGFhJpcfTRRzuUVxrASije9ttv3zUZmKE+KrrL1+Mn5a/dOOjvRoPlbzNBrR7aVi1K/v8bPgdxgKyxaaYbrax0Z6GKzlBzeyPh9NNPd8edNGlSMF3til771PPlRoPf4tBMRWevLOqxRChd7Yqa6nVeqjwbCWpxUD6q1UHbN9rioLkatJ36AITSlTNqjbNHEEJvpUh0QuN8rL766i5OI48qLfRHHCRO2laPbsppAsQBMsfebtCdZdVghbJ6r1sfh0bHLbAm6LpNKKT+GTovzVpZNUjalIfW1KvtG23FWW+99dx2mkwrlK52xfp8qBWlkaC8VJ76fRwakVsbTVWtMqF05Yy9EaK069FCby1GmiRNcdWhstGgfStfmwlq9dJx77rrrmC6Oh3EAbJGHa70A1Zza9XgF8oKzXTos05bmm8hlK52xUZUlBBVCf6dsf5WsGnIq7YCafZDxdcbJqE0tTN6bKVz03DoVYNdj2qyV1C+6nptRG7V1K99tGuLmDp3hpaX0RDsOk+JfKvCp59+6vqPqGVOfatC6ep0EAfImqlTp7qCQ2MfVAnWPGmFsgXr71ClstNdseJqtLpQmtoZ9bTXuWnEyipBEqb4flO6KjqJhKgS7A0VdXYLpandUYWv86syy6hJgmTMREyhUbm1t4301kkoTXVB4q7zlMi3Ktj/QmVGKE2AOEAbYNMzq4mzr6DCWE2+fqFsQcurNAerU5iOp85gofS0M8oXnZvQcMV9hZ7yrLd8Lgd1bNTx+hrsp1058cQT3fmpNadK6CnfesrrcrDHRerkGkpP3bCxHPSqaSuCJn/T8QZinJV2AXGA7NH75fohr7POOtN/2umCenPrWEIzLIbS0+5YK4J62qcOKux1rHbsxFcVva2ieTF0nhpjInWw5/5qnQilp25ozAmdr24gUodbbrnFHUvTzofSAp+DOEBbsO6667oftHpfpwx6hU3HUatDKB11QD3yNQ6FzjN1RWetRXqdNZSWunD88ce786z6CKjZoNccdRy9khpKR13RwGs6bw0qlTLYvCN16xQdG8QB2gINqqQftNDQtCmC3sXX/vUuuWauDKWjLlgFpHEy/vznP0/PgbjBJmLaaKONgmmoE3ozwF6t1EijKYJGzbSWDY3iGEpHXbF+B0JvOqQImjdG+089K24dQBygbdBgMvpha8IkdZqKGWwqXaFCKnT8umET+Wy99dbTcyFesAF09CZFXR/5lLG+B0JDO8cMmghNI35q3xqwKXT8uqNhvXX+GpZaA7HFDNZipE7YAzUkfDuBOEBboRED9QOXPGg+gxhBd2/ap9DAT6Hj1hG1NKy22mruvDfbbDM3hXiMoNEELT877T14zelg597oWBc9heeee65rICSNqBk6bqdgjxI1V4xaIWMEPf60/5nmTAkdF2YEcYC2Qu9Va3Ic+6H3p4Ofhu3V0Lm2LzUxh45ZZ9T8bf0QJGNVx/IPBc0VYSNEqg+F7rpDx6w7NvaA0BDPmlys2aDrc95553X70qBE7777bvCYnYSN6ijUUtDoPDYW9Mjz61//utuPWsY0P03oeNAdxAHaEntGL9Zff33XG7qRoP4MKiy0fSdXckID3vgypspf4wNUDZpe3BcwzT/w6KOPBo/VKeh61KBQyg/NjnryyScXn3322fQc6ztoXhQbbl3oTZjQcToVTSZnebPooou61p0//elP03Ov96CZQm1EWqH5TzrlcVosEAdoWzQugN0tCw2nrAl8NGXzCy+84AoSFdYaXvmpp55y486rgrQOZsIqydD+Ow1Ne6wpty1vNMiQRijUVMkvv/yym+lTd3d6BiwxkHzpEYfFF3r+zl3x56gfjoaitrzRSISazEuPxvQ2y5tvvln87//+rxvTQXE1xbjupjUFu22jQcj0OnJo/52OfreaKMvySqgFYcstt3SvcOqxg+JoGndNBa7Wn2WXXbYrrvpKqN9EaN/QO4gDtD1XXHFF1zPgMpqWN7Rc78LzPLM7EgMNY+wXsH2hpnSNCvmb3/wmuM9OR31x1BF1lllmCeZfiBEjRjhhC+0PZkSP19SCMGTIkGBelpEQa7ZOPUIK7Q/6BnGA2qC7Yr3nramaNQWz3nVXM7EKYU3PPHbsWPdIQgP2hLaHGXnsscfcM2QNvatJqiQTmuZYBa866al1QW+jhLaF7qjFRgNi7bnnnm6CJ3VMVWWnviVqLlfr1xlnnOH6nYS2h77R7KsSX00XrvxUi5hes9SjND3eePXVV4PbQWMgDlBr1GM6tBwaR8Mhq8k3tA6ao47DmkP9aYk46DlT7s+R9SpaHZ51q5NPM+ehbZQHoXXtijqYqT8DM9zF4aSTTnLPkEProDnUp0SvFYbWAeRKS8RBg2ro2ZLecQ6tV4WlZk+/4lJFlnKgE+1fAwr533tLY7tg75E3Kg+asa9Kft90002O0Lrc0HN35QWdy+Kw4ooruvxUx9PQemgMGw1VTeqh9QC5klwcVMnoxzFq1CgnEKE4GnRHcXxx2GqrrXqMHwPt3594x+RFz8j8eO2GCVejrxdVEQflkf5P7TBI0scff9zVGa2O02O3mgcffNDlpeDVwDjY/AvinXfeCcYByJHk4mAVtAlEqELLQRw6nbqJgyapsUJZ0Cmqf+hVNstLdZIMxYHq6FVM//pUp95QPIAcSSoOqmhUIdkjAf2tnu1+HK2z+db1qUpJcezxhpb5rQBqitcy7cvil/en7XXn3VM8kxJ//xbfP5a1Qmi54mq78qMMHUv79tOleL0152ud4voSZccv71/78h+pKH12HJ1DubK3dJT3rXTaNkqvtvO31Tp91zr/PLSt7UPLlQ9ap7+13PJI+7XlfnoHCr1FobQaejsgFA+qoYm//PzstKGkY6PfmZ+fnTARGNSHpOJgPw6rxFTBqHLxKzVVhKqgFE/x9d0qVsVVRVjeXo89FMf278uIxVFFpuWKp/iKZ5Wy7V9xbP+qGP04qhAtDaoItY3258cR2rcdT/GUJv2teP55+oTu3O1ctK9yPB3bj6NjKg1C8a0SF1qmOFbhKw2Ko22Ut7aN0qhltp2WaTur+HUsLRNar7TYvn2BMiGxPNKn4uj/YPtuNS+++KJLgwbc0afQK5mhuNA3GkxHeWhTDgu94haKC9WwCauE5l3QpyZuCsUFyI2k4qCKya+crHL2K0xhFaIqJ1umykmVm31Xpac4qphsmbDKzJcLfbeKzVDlpn3ad/3tV7hlcbAKUMe1OELbaV92PJ2f4llFLbROy8rn6aPtyukx4bB88M9Ny8rnICxfLN22jaXH9uvnraWvLA7leJYHti+t03f/vLRNOU3arvx/aiXHHHOMS6dN4GSDGd13333B+NA7GolP+bfUUku5TzFo0CA3LkEoPvTOww8/PEN+qg+OPidMmBCMD5AbycShJ0lQRaNKyl9WRRwsju5qtW/DKk47jomDVeyGJMGv4PR3b+JQXm/Y8UwoVPn66RRWwfZWeSq9fj5oH1bp+2mwNNtxtU///IVfeZfFQetCd/9Kd1kcbB+G9qF9WYtHSBysFUbHsHgDjQ3Zq+F99fmd73zHfTKmQ+No7AblnQbS0u9Bf5tIMKZDc2hYdP+61O9OnxoyPRQfIDeSiYPdrQpVSj5aZhWvqCIOVkH1hFXSJg62ndGoOPiVsY/drVu8cgVs+GkKYZWw8sGOrX3ruDpXxdHftg/Lo56wNJTFQX+H0qFjlMWhLBhVxEGU/zfKNzt+q9HokUqD5rAYM2aM+/uiiy5ynxrTIbQN9Iw67SnvNAuprkf9bdOQM6ZDc1h/EYmXPjUypyZq0988roB2IIk4qIJRRaSKSRVZGf1ArHIUVcTBhEAVkuKF8OPZdkaj4lCOX45nlWmz4iBMEhTP0qL0K++0f+3DWk7suz5D5y4Uz/LXKu6QEAidWyxxMBTP8l/7C8VpBZqZUBNb2dz9N9xwQ/HAAw90/IyNzaK5QPQ5fPhwl5+vv/66mzBs6tSp3eJC32jCNV2PNo6Dxht5++23i5tvvjkYHyA3koiDX8mF1tsdqlWKIXFQHF8cbJ/lSkt37KoErQWjEXHw918WB6vAbb2hSt5Pe3/EwY6htFilbeep8/fFxiptX7hsuba3dJfFQWkrn4fSrmUxxEHblP/P9v/0W5UGAl8cQuuhMXQ9Kj8lDqH10Bj2W99jjz2C6wFyJYk4WGXli4BPWQKsstN328YqVVU+tkwVvbAKSRWbjqUCzd9O+9LfPmVx8Pev72VxUOWq737lqrjaxq+8td6PY2jbvsRB+1M87bNcOYe2V/oV19KouLbMRKEsDnZeiqfjCaVXy/x0ax9VxUH7sv0rL7Stn4/ar5bZ/2SgQBzigjjERdPCKz/32muv4HqAXIkuDlbZlCuhMhIAVS76WxWM/rZKScusUhVWgWrfVngJbaNKyiotUVUc/P1LWCzdVikLVZiWLn0KVZR+hajj+xWwoW36Egdh+VDep7b3z0sojo6vdYbOy09zWRyE9qPjaLmOpTzSdn66bbl9F5YnJg7C0iaUHmH7NrQvf5uBAnGIC+IQF12Xyk867UK7kaTFQXfrfkUYQuvL8VTB+RWeLbPHAoa+a3noGLbfqst1PFuuz9A+VQk2erye9lUmtH1P+zS0Tmkq55W/3o5dToN9V2Xvt5xom9Ax/X35y3Rsf7nyx9JUjj9QIA5xQRzicv3117v8HDduXHA9QK4kEQfIB3uU4S9TBa8Cq0qLSDuDOMQFcYjLdddd5/JTb6yE1gPkCuJQc9QSoMJJhb4eRehRg0Si/MiljiAOcUEc4nLttde6/NQ8IKH1ALmCOHQAEgT141Drg+RB/SDqLg0CcYgL4hCXyZMnu/w84IADgusBcgVxgNqCOMQFcYjL1Vdf7fLzwAMPDK4HyBXEAWoL4hAXxCEuGkRL+XnwwQcH1wPkCuIAtQVxiAviEJcrr7zS5echhxwSXA+QK4gD1BbEIS6IQ1wuv/xyl5+HHnpocD1AriAOUFsQh7ggDnGxwdoOP/zw4HqAXEEcoLYgDnFBHOJis4weccQRwfUAuYI4QG1BHOKCOMTl4osvdvk5fvz44HqAXEEcIGuefPLJ4rzzzmuKjTbayBXMe+65Z3A9NMawYcNcfk6aNCm4Hhpjp512cvm56aabBtf3BdNww0CBOEDWWGUFAN3pab4agJQgDpA1Q4cOdQXk0UcfDQDTWXHFFd3vYsqUKcHfDUBKEAfIGhOHzz77rCAQCJ+HLbfcEnGAAQNxgKxBHAiE7gFxgIEEcYCsQRwIhO4BcYCBBHGArEEcCITuAXGAgQRxgKxBHAiE7gFxgIEEcYCsQRwIhO4BcYCBBHGArEEcCITuAXGAgQRxgKxBHMLhpz/9qcuXZZZZZvqS7uH9998vnn766enfPg/333+/I1XQvv1jbrPNNr2mkdBcQBxgIEEcIGsQh3BQhbzJJpu4vHnttdemL50xqMI+5JBDpn8rXLzBgwcXV1555fQlcYOEQfuX1Fg49thjXVoJcQPiAAMJ4gBZgzh0D2pJsApacrDbbrtNXzNjUJyyOCgvU4qD9u+LAyFNQBxgIEEcIGsQh+5Bd/HKE4mAxECCUG510HLFsVaHhx56yAmGlqmlQvuwIBE566yz3HK1DvjrLGgfOoakQPspx5M0hPav/Zb3p/1omeJpm7LIaF86nqVLxxKphKcdQ1VxuPPOO4uxY8cW1157bXA9QDMgDpA1iEP38NWvfrWr+V+Vq/JHFawFLbNKXHH19y9/+Uu3jVXsqpgtruJIPqzC199aZsGOofUWT9gyBVX2tkzHsf3rb7+Pg/pA2P4VR1g6LSiOfzzF0XotK0tIp4bexOH5558vjjrqqGL55Zd3cYRm4CzHA2gWxAGyBnGYMVil6j8OUMWsCrYcrNK1EHpUoYpYy1TxWzBRkHD437U//W3BRMRaO0KPKnxxMEmREPj7sXOydPni4Mfr6Tw7MZTF4YMPPihOO+20YoMNNnDLjbXWWstNg/7OO+8Ub775Zu0olxfQGhAHyBrEYcagyrxcgavCVR6V+xZUEQdVxqrcy8GOo6BjaTt/Xwo6npabdPQlDqH1FkwoFMoiYcFaJwhfiINamvQoYrbZZnPfO41ZZ521WHLJJYtx48YVt912W7AMgfggDpA1iMMXQRW4KnOhCtmwRwTWQmChL3Gw78L266PlOqaJQ/kxgYmDKnqFvsTBBMdEww9Kp46pEGpVUbDWEcIX4qC8uuSSS4rRo0e778Zcc81VLLbYYsUqq6zipuBeaqmliuHDh9eKQYMGFQsuuOAM573vvvsWH3/8cbAsgXggDpA1iMMXwSpeVcaSBB/rA2CPDRRUEfcmDiYE2l6VdAg/Xn/FweKHxEHxrJ+DiYPt1wLi8EUI9XF44YUXiokTJzpR0DpjzJgxxS233DLD76pOvPrqq8U555zTJRE6/zqfbw4gDpA1iMMXQS0LVgmXg1W2fifJvsRBwVoXyiGFONh+/DRakDRIYBQQh75Db50jxV133eXyU3flirfLLrsE49WJ3/72t8V2223nzlc888wzwXjQfxAHyBrE4fNglXK58vaDKmhfAvS39RtQCFXcVhn7lb0qbG1rx6oqDiYm/v59cVCQIGjffsuI4vv7QRz6Dn2Jg/Hhhx8Wl156afHpp58G19eRCRMmuLz5xje+EVwP/QdxgKxBHD4PVmmGmvktWOdBa1HQHae++9up0tZ3vxVAlbuW2eMCEw6tszha35c4KF55/2VxkDDou+LoGPaIxd834tB3qCoOncraa6/t8uekk04Krof+gThA1iAOnwdV0n6rQCio4pY0+BWuttEdvd3hW5zy4wLFU8UsynJi25SXa59arvUWbJntX2kJpVtxdKzQfu14/n4VFE/LCYhDX2jAM+XPvPPOG1wP/QNxgKxBHAiE7gFx6JvNN9/c5REdJeODOEDWIA4EQveAOPTNmWee6fJILVuh9dA8iANkDeJAIHQPiEPfXH/99S6P9tlnn+B6aB7EAbIGcSAQugfEoW/uuOMOl0c77bRTcD00D+IAWYM4EAjdA+LQN4hDOhAHyBrEgUDoHhCHvkEc0oE4QNYgDgRC94A49A3ikA7EAbIGcSAQugfEoW8Qh3QgDpA1Jg4bbrghAExnwekTOiEOPYM4pANxgKzZdNNN3Y8fAGZE8vD73/8++LsBxCEliANkj6YLbpZx48YFl0PjPPXUU8W5554bXAeNM3ny5OK+++4LrqvCe++9F/y9wOcgDulAHKC2qKJbdtllg+ugcS655JJizJgxwXXQOJJaJmFKh4aaRhzSgDhAbRk/frwrOG6//fbgemiM0aNHu/x8++23g+uhMeabb75i1VVXDa6D/nPVVVe563XNNdcMrofmQRygtswxxxwUHJFQ07jyUjD2f/+54ooruvJTMzmG4kD/uPrqq13+HnTQQcH10DyIA9SSn/3sZ10F8yyzzMLz4H4yceLErvxcY401gnGgOjZzo9hvv/2CcaB/IA7pQBygluywww5dBbO48MILg/GgGiuuuOIM+fnYY48F40HfTJ06dYa8HDJkSDAe9A/EIR2IA9SOadOmFV/60peKmWeeuatwHjVqVDAu9M1dd93l8nDuuefuys/DDz88GBf65oQTTujKx0GDBrnPa6+9NhgXmgdxSAfiALVDrwyqwFhppZXc58ILL+w+n3322WB86J1dd93V5d+wYcPcp1h88cWDcaFv1CHS8tEGctJIkKG40DyIQzoQB6gdG2ywgSswtthiC/e59dZbu88JEyYE40PPfPjhh12dTEeMGOE+NXKhPm+++ebgNtAz999/v8u7kSNHuk/J2ODBg93fr7/+enAbaA7EIR2IA9SKJ5980hUWyy23XLH33nu7v0877TT3yZgOjaOxG5R3Gr9h5ZVXdn+feuqpXctC20DPaOwG5d2kSZPc5yKLLNK1jDEd4oI4pANxgNpx9NFHu3e4x44d6wqOG264wbU2XHDBBcH40Dv7779/8eCDDxbLLLOMy8+XXnqp2HHHHYsbb7wxGB965oEHHnAdd1955RWXlwsssEDxxBNPFHvuuWfx7rvvBreB5kAc0oE4QG3ZZ599usQhtB4aw5rXaVLvPx999JHLSz2qCK2H/oM4pANxgNqCOMQFcYgH4pAexCEdiAPUFsQhLohDPBCH9CAO6UAcoLYgDnFBHOKBOKQHcUgH4gC1BXGIC+IQD8QhPYhDOhAHqC2IQ1wQh3ggDulBHNKBOEBtQRzigjjEA3FID+KQDsQBagviEBfEIR6IQ3oQh3QgDlBbEIe4IA7xQBzSgzikA3GA2oI4xAVxiAfikB7EIR2IA9QWxCEuiEM8EIf0IA7pQBygtiAOcUEc4oE4pAdxSAfiALUFcYgL4hAPxCE9iEM6EAeoLYhDXBCHeCAO6UEc0oE4QG1BHOKCOMQDcUgP4pAOxAFqC+IQF8QhHohDehCHdCAOUFsQh7ggDvFAHNKDOKQDcYDagjjEBXGIB+KQHsQhHYgD1BbEIS6IQzwQh/QgDulAHKC2IA5xQRzigTikB3FIB+IAtQVxiAviEA/EIT2IQzoQB6gtiENcEId4IA7pQRzSgThA1tx+++3F9773vaZYffXVXcGxzTbbBNdX4Y033gimq1256KKLgudZhaFDh7r83G+//YLr+2LChAnBNLUzxx13XPBc++KII45weTnvvPMG11fh7LPPDqYJPgdxSAfiAFljldVAseOOOwbT1Y689NJLwXNsJWeddVYwbe3I5MmTg+fYSqZMmRJMGyAOKUEcIGtMHEYdeen/c1nDrLPXScHlfbHamMPccesoDvMttlzwnPti3XGnNZ2fy2ywtTt2HcVh2Bqjg+fcF8rLZvNzyPKruWMjDj2DOKQDcYCsMXHY+77Pir3vL1rGFqff445bR3FY5MvrBM85JWvufpw7dh3FYcS2BwXPOSVLrbelOzbi0DOIQzoQB8gaxCEeiENcEIe8QRzSgThA1iAO8UAc4oI4tIZnn322KU466SSXRzvvvHNwfRVC6QHEATIHcYgH4hAXxCE9hx56qDvPgWLzzTcPpqvTQRwgaxCHeCAOcUEc0rPtttu681z2y6sXI9bepCG+suZGxeAhQ4sVRqwZXN8bw9fc2B13ySWXDKar00EcIGsQh3ggDnFBHNJj4nDSNQ8Wt75UtIybnvvUHRdxCIM4QNYgDvFAHOKCOKQHccgTxAGyBnGIB+IQF8QhPYhDniAOkDWIQzwQh7ggDulBHPIEcYCsQRzigTjEBXFID+KQJ4gDZA3iEA/EIS6IQ3oQhzxBHCBrEId4IA5xQRzSgzjkCeIAWYM4xANxiAvikB7EIU8QB8gaxCEeiENcEIf0IA55gjhA1iAO8UAc4oI4pAdxyBPEAbIGcYgH4hAXxCE9iEOeIA6QNYhDPBCHuCAO6UEc8gRxgKxBHOKBOMQFcUgP4pAniANkDeIQD8QhLohDehCHPEEcIGsQh3ggDnFBHNKDOOQJ4gBZgzjEA3GIC+KQHsQhTxAHyBrEIR6IQ1wQh/QgDnmCOEDWIA7xQBzigjikB3HIE8QBsgZxiAfiEBfEIT2IQ54gDpA1iEM8EIe4IA7pQRzyBHGArEEc4oE4xAVxSA/ikCeIA2QN4hAPxCEuiEN6EIc8QRwgaxCHeCAOcUEc0oM45AniAFmDOMQDcYgL4pAexCFPEAfIGsQhHohDXBCH9CAOeYI4QNYgDvFAHOKCOKQHccgTxAGyBnGIB+IQF8QhPYhDniAOkDWIQzwQh7ggDulBHPIEcYCsQRzigTjEBXFID+KQJ4gDZA3iEA/EIS6IQ3oQhzxBHCBrEId4IA5xQRzSgzjkCeIAWYM4xANxiAvikB7EIU8QB8gaxCEeiENcEIf0IA55gjhA1iAO8UAc4oI4pAdxyBPEAbIGcYgH4hAXxCE9iEOeIA6QNYhDPBCHuCAO6UEc8gRxgKxBHOKBOMQFcUgP4pAniANkDeIQD8QhLohDehCHPEEcIGsQh3ggDnFBHNKDOOQJ4gBZgzjEA3GIC+KQHsQhTxAHyBrEIR6IQ1wQh/QgDnmCOEDWIA7xQBzigjikB3HIE8QBsgZxiAfiEBfEIT2IQ54gDpA1iEM8EIe4IA7pQRzyBHGArEEc4oE4xAVxSA/ikCeIA2QN4hAPxCEuiEN6EIc8QRwgaxCHeCAOcUEc0oM45AniAFmDOMQDcYgL4pAexCFPEAfIGsQhHohDXBCH9CAOeYI4QNYgDvFAHOKCOKQHccgTxAGyBnGIB+IQF8QhPYhDniAOkDWIQzwQh7ggDulBHPIEcYCsQRzigTjEBXFID+KQJ4gDZA3iEA/EIS6IQ3oQhzxBHCBrEId4IA5xQRzSgzjkCeIAWYM4xANxiAvikB7EIU8QB8gaxCEeiENcEIf0IA55gjhA1pg4LLzymi1l/iVWcsetozjMNmju4DmnZO6FFnfHrqM4zLnAIsFzTsns88zvjt0p4jBs2ZWLFUas2TKWH76GOy7iEAZxgKzZYost3A94oJg4cWIwXe3KCiusEDzPVnHrrbcG09WOPPLII8FzbBWS6j/84Q/BtNWF8ePHB8+9VWyzzTbBdHU6iANkz9SpUweMUHrandB5toLXX389mJ52RhV36FxbwSeffBJMU90InXurCKUHEAcAAABoAMQBAAAAKoM4AAAAQGUQBwAAAKgM4gAAAACVQRwAAACgMogDAAAAVAZxAAAAgMogDgAAAFAZxAEAAAAqgzgAAABAZRAHAAAAqAziAAAAAJVBHAA6hNNPP71Yeumli2nTpgXXAwBUAXEA6BAkDjPNNFPxyiuvBNcDAFQBcQDoEBAHAIgB4gDQIYTE4d5773XLH3/88a5lepRx2WWXFQcddFAxceLE4qabbupaJxS/vEzYvvzv2n7s2LHuE2GBquh6GTlypHu0ttVWW7nrU9/tGtJ1pu/lx2661oS/TNfhqFGj3L6Eruvyeu3LjqG4xx57rPs7dJ0rPUpfeXkngTgAdAhlcVCBOXjwYFeQWgGsdSowFU8FqNDffmGrglnb2XdDcVWo6m87lvZl8fU9VBAD+Oja0fVi0qprShW+rh8TXLu+yuKga1DXnH3X9tqXlmt/EmLty4+j34H2peV2zWuZfbd4QsfnOkYcADoGXxxC0iCsktd6W1YuLFX4lgtPi6N1+q5C1yTCCBXEAD52HZXv6HUtaXmj4qC/dd35cbSNttU+9N3EoXy96reh5X5LmdKl30f5uJ0G4gDQIVhhq8o9JA1WoJabekVZBPTdj1cuULVedPqdGTSGrpdyZe0vb0QcTEJ03epvH/96NnEw6TXs92CCIbSd3/rWqSAOAB2CFbZG+Q7LCloJgFX8hpYJi6vC0/+uOP7+rDD2j1UumAHK2DVaFge7NvXpx+tNHMrXYBlds368kORqfxbP5EXxy/E6DcQBoEOwwlaVfuhxgxWgquQVt4xf8atgV1wts0I9VPBqvd/HgUcV0Bt2XZbFwa7NZsRB+1S8EH68kBD46dHvwn8M0skgDgAdghW2ViirEFSFbt9VkGp96FGFpMAKbcNaGfSYwu7KhPYTiq+4/vEByviVvb9c15iW2zUVEgxdd7oOrXK367ksq1qua9xEtzdxUFz9RvTb0b7LfS86FcQBoEMoi4M+9V0VusVRIauC0q/0VcAqXrnQVOGtuKEC1S/ADcVRfBXG/nIAHxNauwb1qe+6Bv1l/rVrMqB4/nWnZYpnIqJ49pjNRKE3cRB+ixnS+zmIA0CHUBYHoUJUy+zuS+tU6WuZCkwrxCUU5Qpf3xUvVKCabGhbfz90LIO+0HWl60XXj65FXTfWWuULrV27hq5Ru9b8fZk8CO1L+KLblziUJQUQB4COQQWgCsyyAGiZpKK8TAWuPrVdeRtDhXW5KdhQQaztVeCqkO+pYAYIoetFAqprz0TUFwchYVVrgi1X3HIcW659aJ+hVoPernHF17FNrgFxAIAmUUGru7fy82iA2PQkDq1A8qvrvCex6EQQBwBoCN21SRbU0qCmZApUSM1AiIOEwR6ZlFvkOh3EAQAaQuKgwlTSwOMHaAUSBj3uCj1mSIXkWI/ZkIbuIA4AAABQGcQBAAAAKoM4AAAAQGUQBwAAAKgM4gAAAACVQRwAAACgMogDAAAAVAZxAAAAgMogDgAAAFAZxAEAAAAqgzgAAABAZRAHAAAAqAziAAAAAJVBHAAAAKAyiAMAAABUBnEAAACAyiAOAAAAUBnEAQAAACqDOAAAAEBlEAcAAACoDOIAAAAAlUEcAAAAoDKIAwAAAFQGcQAAAIDKIA4AAABQGcQBAAAAKoM4AAAAQGUQBwAAAKgM4gAAAACVQRwAAACgMogDAAAAVAZxAAAAgMogDgAAAFAZxAEAAAAq8rfi/wBLLq4UAPn/rAAAAABJRU5ErkJggg=="}},"cell_type":"markdown","metadata":{"graffitiCellId":"id_he9lhbo","jupyter":{},"tags":[],"slideshow":{"slide_type":"slide"},"id":"DC31ED3C042A4C68936B8C9C9493C4D4","mdEditEnable":false},"source":"# 注意力机制\n在“编码器—解码器（seq2seq）”⼀节⾥，解码器在各个时间步依赖相同的背景变量（context vector）来获取输⼊序列信息。当编码器为循环神经⽹络时，背景变量来⾃它最终时间步的隐藏状态。将源序列输入信息以循环单位状态编码，然后将其传递给解码器以生成目标序列。然而这种结构存在着问题，尤其是RNN机制实际中存在长程梯度消失的问题，对于较长的句子，我们很难寄希望于将输入的序列转化为定长的向量而保存所有的有效信息，所以随着所需翻译句子的长度的增加，这种结构的效果会显著下降。\n\n与此同时，解码的目标词语可能只与原输入的部分词语有关，而并不是与所有的输入有关。例如，当把“Hello world”翻译成“Bonjour le monde”时，“Hello”映射成“Bonjour”，“world”映射成“monde”。在seq2seq模型中，解码器只能隐式地从编码器的最终状态中选择相应的信息。然而，注意力机制可以将这种选择过程显式地建模。\n\n![Image Name](https://cdn.kesci.com/upload/image/q5km4dwgf9.PNG?imageView2/0/w/960/h/960)\n\n## 注意力机制框架\n\nAttention 是一种通用的带权池化方法，输入由两部分构成：询问（query）和键值对（key-value pairs）。$𝐤_𝑖∈ℝ^{𝑑_𝑘}, 𝐯_𝑖∈ℝ^{𝑑_𝑣}$. Query  $𝐪∈ℝ^{𝑑_𝑞}$ , attention layer得到输出与value的维度一致 $𝐨∈ℝ^{𝑑_𝑣}$. 对于一个query来说，attention layer 会与每一个key计算注意力分数并进行权重的归一化，输出的向量$o$则是value的加权求和，而每个key计算的权重与value一一对应。\n\n为了计算输出，我们首先假设有一个函数$\\alpha$ 用于计算query和key的相似性，然后可以计算所有的 attention scores $a_1, \\ldots, a_n$ by\n\n\n$$\na_i = \\alpha(\\mathbf q, \\mathbf k_i).\n$$\n\n\n我们使用 softmax函数 获得注意力权重：\n\n\n$$\nb_1, \\ldots, b_n = \\textrm{softmax}(a_1, \\ldots, a_n).\n$$\n\n\n最终的输出就是value的加权求和：\n\n\n$$\n\\mathbf o = \\sum_{i=1}^n b_i \\mathbf v_i.\n$$\n\n\n![Image Name](https://cdn.kesci.com/upload/image/q5km4ooyu2.PNG?imageView2/0/w/960/h/960)\n\n不同的attetion layer的区别在于score函数的选择，在本节的其余部分，我们将讨论两个常用的注意层 Dot-product Attention 和 Multilayer Perceptron Attention；随后我们将实现一个引入attention的seq2seq模型并在英法翻译语料上进行训练与测试。"},{"cell_type":"code","execution_count":4,"metadata":{"graffitiCellId":"id_lyizjfl","jupyter":{},"tags":[],"slideshow":{"slide_type":"slide"},"id":"2D4703A032B741F68676C00CF5B06CAF","collapsed":false,"scrolled":false},"outputs":[],"source":"import math\nimport torch \nimport torch.nn as nn"},{"cell_type":"code","execution_count":5,"metadata":{"_cell_guid":"79c7e3d0-c299-4dcb-8224-4455121ee9b0","_uuid":"d629ff2d2480ee46fbb7e2d37f6b5fab8052498a","graffitiCellId":"id_ucnczrs","scrolled":false,"jupyter":{},"tags":[],"slideshow":{"slide_type":"slide"},"id":"60B68D50106A4FBD9CA3C2B863146405","collapsed":false},"outputs":[{"output_type":"stream","text":"dirs []\nfiles ['_about.txt', 'fra.txt']\n","name":"stdout"}],"source":"import os\ndef file_name_walk(file_dir):\n    for root, dirs, files in os.walk(file_dir):\n#         print(\"root\", root)  # 当前目录路径\n         print(\"dirs\", dirs)  # 当前路径下所有子目录\n         print(\"files\", files)  # 当前路径下所有非目录子文件\n\nfile_name_walk(\"/home/kesci/input/fraeng6506\")"},{"cell_type":"markdown","metadata":{"graffitiCellId":"id_bimjldk","jupyter":{},"tags":[],"slideshow":{"slide_type":"slide"},"id":"4C277595040B4A0689320409E624C366","mdEditEnable":false},"source":"### Softmax屏蔽\n在深入研究实现之前，我们首先介绍softmax操作符的一个屏蔽操作。"},{"cell_type":"code","execution_count":6,"metadata":{"graffitiCellId":"id_39i6msf","jupyter":{},"tags":[],"slideshow":{"slide_type":"slide"},"id":"CD9DB3F8099F49FB886E0B5029173F44","collapsed":false,"scrolled":false},"outputs":[],"source":"def SequenceMask(X, X_len,value=-1e6):\n    maxlen = X.size(1)\n    #print(X.size(),torch.arange((maxlen),dtype=torch.float)[None, :],'\\n',X_len[:, None] )\n    mask = torch.arange((maxlen),dtype=torch.float)[None, :] >= X_len[:, None]   \n    #print(mask)\n    X[mask]=value\n    return X"},{"cell_type":"code","execution_count":7,"metadata":{"graffitiCellId":"id_mlvtr3b","jupyter":{},"tags":[],"slideshow":{"slide_type":"slide"},"id":"B0BE96AC54A940878423D3C201ED8477","collapsed":false,"scrolled":false},"outputs":[],"source":"def masked_softmax(X, valid_length):\n    # X: 3-D tensor, valid_length: 1-D or 2-D tensor\n    softmax = nn.Softmax(dim=-1)\n    if valid_length is None:\n        return softmax(X)\n    else:\n        shape = X.shape\n        if valid_length.dim() == 1:\n            try:\n                valid_length = torch.FloatTensor(valid_length.numpy().repeat(shape[1], axis=0))#[2,2,3,3]\n            except:\n                valid_length = torch.FloatTensor(valid_length.cpu().numpy().repeat(shape[1], axis=0))#[2,2,3,3]\n        else:\n            valid_length = valid_length.reshape((-1,))\n        # fill masked elements with a large negative, whose exp is 0\n        X = SequenceMask(X.reshape((-1, shape[-1])), valid_length)\n \n        return softmax(X).reshape(shape)"},{"cell_type":"code","execution_count":8,"metadata":{"graffitiCellId":"id_gjfsirw","jupyter":{},"tags":[],"slideshow":{"slide_type":"slide"},"id":"2DC35416E322461FB2B9FDD6E0379A02","collapsed":false,"scrolled":false},"outputs":[{"output_type":"execute_result","metadata":{},"data":{"text/plain":"tensor([[[0.5423, 0.4577, 0.0000, 0.0000],\n         [0.5290, 0.4710, 0.0000, 0.0000]],\n\n        [[0.2969, 0.2966, 0.4065, 0.0000],\n         [0.3607, 0.2203, 0.4190, 0.0000]]])"},"transient":{},"execution_count":8}],"source":"masked_softmax(torch.rand((2,2,4),dtype=torch.float), torch.FloatTensor([2,3]))"},{"cell_type":"markdown","metadata":{"graffitiCellId":"id_s7xh1br","jupyter":{},"tags":[],"slideshow":{"slide_type":"slide"},"id":"6E17FB479C8E461185DC7E64B3B51A5D","mdEditEnable":false},"source":"**超出2维矩阵的乘法** \n\n$X$ 和 $Y$ 是维度分别为$(b,n,m)$ 和$(b, m, k)$的张量，进行 $b$ 次二维矩阵乘法后得到 $Z$, 维度为 $(b, n, k)$。\n\n\n$$\n Z[i,:,:] = dot(X[i,:,:], Y[i,:,:])\\qquad for\\ i= 1,…,n\\ .\n$$\n"},{"cell_type":"code","execution_count":9,"metadata":{"graffitiCellId":"id_c01uiu4","jupyter":{},"tags":[],"slideshow":{"slide_type":"slide"},"id":"55C627C9AD274B3AA4366F554A7A33C8","collapsed":false,"scrolled":false},"outputs":[{"output_type":"execute_result","metadata":{},"data":{"text/plain":"tensor([[[3., 3.]],\n\n        [[3., 3.]]])"},"transient":{},"execution_count":9}],"source":"torch.bmm(torch.ones((2,1,3), dtype = torch.float), torch.ones((2,3,2), dtype = torch.float))"},{"cell_type":"markdown","metadata":{"graffitiCellId":"id_5l05rc9","jupyter":{},"tags":[],"slideshow":{"slide_type":"slide"},"id":"BB8410C97D6143F4885BFA00C8564260","mdEditEnable":false},"source":"# 点积注意力\nThe dot product 假设query和keys有相同的维度, 即 $\\forall i, 𝐪,𝐤_𝑖 ∈ ℝ_𝑑 $. 通过计算query和key转置的乘积来计算attention score,通常还会除去 $\\sqrt{d}$ 减少计算出来的score对维度𝑑的依赖性，如下\n\n\n$$\n𝛼(𝐪,𝐤)=⟨𝐪,𝐤⟩/ \\sqrt{d} \n$$\n\n假设 $ 𝐐∈ℝ^{𝑚×𝑑}$ 有 $m$ 个query，$𝐊∈ℝ^{𝑛×𝑑}$ 有 $n$ 个keys. 我们可以通过矩阵运算的方式计算所有 $mn$ 个score：\n\n\n$$\n𝛼(𝐐,𝐊)=𝐐𝐊^𝑇/\\sqrt{d}\n$$\n \n现在让我们实现这个层，它支持一批查询和键值对。此外，它支持作为正则化随机删除一些注意力权重."},{"cell_type":"code","execution_count":10,"metadata":{"graffitiCellId":"id_qpr6dwm","jupyter":{},"tags":[],"slideshow":{"slide_type":"slide"},"id":"3CB825EFC83C4F7892F0ED1D4038FB49","collapsed":false,"scrolled":false},"outputs":[],"source":"# Save to the d2l package.\nclass DotProductAttention(nn.Module): \n    def __init__(self, dropout, **kwargs):\n        super(DotProductAttention, self).__init__(**kwargs)\n        self.dropout = nn.Dropout(dropout)\n\n    # query: (batch_size, #queries, d)\n    # key: (batch_size, #kv_pairs, d)\n    # value: (batch_size, #kv_pairs, dim_v)\n    # valid_length: either (batch_size, ) or (batch_size, xx)\n    def forward(self, query, key, value, valid_length=None):\n        d = query.shape[-1]\n        # set transpose_b=True to swap the last two dimensions of key\n        \n        scores = torch.bmm(query, key.transpose(1,2)) / math.sqrt(d)\n        attention_weights = self.dropout(masked_softmax(scores, valid_length))\n        print(\"attention_weight\\n\",attention_weights)\n        return torch.bmm(attention_weights, value)"},{"cell_type":"markdown","metadata":{"graffitiCellId":"id_sy1n984","jupyter":{},"tags":[],"slideshow":{"slide_type":"slide"},"id":"CD3484D8F62449C08842C768228EA444","mdEditEnable":false},"source":"### 测试\n现在我们创建了两个批，每个批有一个query和10个key-values对。我们通过valid_length指定，对于第一批，我们只关注前2个键-值对，而对于第二批，我们将检查前6个键-值对。因此，尽管这两个批处理具有相同的查询和键值对，但我们获得的输出是不同的。"},{"cell_type":"code","execution_count":11,"metadata":{"graffitiCellId":"id_rt2wg6w","jupyter":{},"tags":[],"slideshow":{"slide_type":"slide"},"id":"DF454887CCD745858E6505AF13119489","collapsed":false,"scrolled":false},"outputs":[{"output_type":"stream","text":"attention_weight\n tensor([[[0.5000, 0.5000, 0.0000, 0.0000, 0.0000, 0.0000, 0.0000, 0.0000,\n          0.0000, 0.0000]],\n\n        [[0.1667, 0.1667, 0.1667, 0.1667, 0.1667, 0.1667, 0.0000, 0.0000,\n          0.0000, 0.0000]]])\n","name":"stdout"},{"output_type":"execute_result","metadata":{},"data":{"text/plain":"tensor([[[ 2.0000,  3.0000,  4.0000,  5.0000]],\n\n        [[10.0000, 11.0000, 12.0000, 13.0000]]])"},"transient":{},"execution_count":11}],"source":"atten = DotProductAttention(dropout=0)\n\nkeys = torch.ones((2,10,2),dtype=torch.float)\nvalues = torch.arange((40), dtype=torch.float).view(1,10,4).repeat(2,1,1)\natten(torch.ones((2,1,2),dtype=torch.float), keys, values, torch.FloatTensor([2, 6]))"},{"cell_type":"markdown","metadata":{"graffitiCellId":"id_vqmu7c7","jupyter":{},"tags":[],"slideshow":{"slide_type":"slide"},"id":"AD0E0FB0D571413E8DE3796462784DD4","mdEditEnable":false},"source":"# 多层感知机注意力\n在多层感知器中，我们首先将 query and keys 投影到  $ℝ^ℎ$ .为了更具体，我们将可以学习的参数做如下映射 \n$𝐖_𝑘∈ℝ^{ℎ×𝑑_𝑘}$ ,  $𝐖_𝑞∈ℝ^{ℎ×𝑑_𝑞}$ , and  $𝐯∈ℝ^h$ . 将score函数定义\n$$\n𝛼(𝐤,𝐪)=𝐯^𝑇tanh(𝐖_𝑘𝐤+𝐖_𝑞𝐪)\n$$\n. \n然后将key 和 value 在特征的维度上合并（concatenate），然后送至 a single hidden layer perceptron 这层中 hidden layer 为  ℎ  and 输出的size为 1 .隐层激活函数为tanh，无偏置."},{"cell_type":"code","execution_count":12,"metadata":{"graffitiCellId":"id_2pmvrcv","jupyter":{},"tags":[],"slideshow":{"slide_type":"slide"},"id":"047BD3F3C6DF46AC8C0FBAFF61F34B75","collapsed":false,"scrolled":false},"outputs":[],"source":"# Save to the d2l package.\nclass MLPAttention(nn.Module):  \n    def __init__(self, units,ipt_dim,dropout, **kwargs):\n        super(MLPAttention, self).__init__(**kwargs)\n        # Use flatten=True to keep query's and key's 3-D shapes.\n        self.W_k = nn.Linear(ipt_dim, units, bias=False)\n        self.W_q = nn.Linear(ipt_dim, units, bias=False)\n        self.v = nn.Linear(units, 1, bias=False)\n        self.dropout = nn.Dropout(dropout)\n\n    def forward(self, query, key, value, valid_length):\n        query, key = self.W_k(query), self.W_q(key)\n        #print(\"size\",query.size(),key.size())\n        # expand query to (batch_size, #querys, 1, units), and key to\n        # (batch_size, 1, #kv_pairs, units). Then plus them with broadcast.\n        features = query.unsqueeze(2) + key.unsqueeze(1)\n        #print(\"features:\",features.size())  #--------------开启\n        scores = self.v(features).squeeze(-1) \n        attention_weights = self.dropout(masked_softmax(scores, valid_length))\n        return torch.bmm(attention_weights, value)"},{"cell_type":"markdown","metadata":{"graffitiCellId":"id_o2cxg9q","jupyter":{},"tags":[],"slideshow":{"slide_type":"slide"},"id":"40A0160DE63247F881553CC224705D7A","mdEditEnable":false},"source":"### 测试\n尽管MLPAttention包含一个额外的MLP模型，但如果给定相同的输入和相同的键，我们将获得与DotProductAttention相同的输出"},{"cell_type":"code","execution_count":13,"metadata":{"graffitiCellId":"id_ji2koe2","jupyter":{},"tags":[],"slideshow":{"slide_type":"slide"},"id":"A399CB39F56241048E63FFD6FD3E3AE9","collapsed":false,"scrolled":false},"outputs":[{"output_type":"execute_result","metadata":{},"data":{"text/plain":"tensor([[[ 2.0000,  3.0000,  4.0000,  5.0000]],\n\n        [[10.0000, 11.0000, 12.0000, 13.0000]]], grad_fn=<BmmBackward>)"},"transient":{},"execution_count":13}],"source":"atten = MLPAttention(ipt_dim=2,units = 8, dropout=0)\natten(torch.ones((2,1,2), dtype = torch.float), keys, values, torch.FloatTensor([2, 6]))"},{"cell_type":"markdown","metadata":{"graffitiCellId":"id_qqem3ml","jupyter":{},"tags":[],"slideshow":{"slide_type":"slide"},"id":"C4CD9496F89B418F9420B4ECD6DE4A60","mdEditEnable":false},"source":"# 总结\n\n- 注意力层显式地选择相关的信息。\n- 注意层的内存由键-值对组成，因此它的输出接近于键类似于查询的值。"},{"attachments":{"attention2.2.PNG":{"image/png":"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"},"attetion2.PNG":{"image/png":"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"}},"cell_type":"markdown","metadata":{"graffitiCellId":"id_1b80ob1","jupyter":{},"tags":[],"slideshow":{"slide_type":"slide"},"id":"28BBCA401DCC416385D179EC8A706DE6","mdEditEnable":false},"source":"# 引入注意力机制的Seq2seq模型\n\n本节中将注意机制添加到sequence to sequence 模型中，以显式地使用权重聚合states。下图展示encoding 和decoding的模型结构，在时间步为t的时候。此刻attention layer保存着encodering看到的所有信息——即encoding的每一步输出。在decoding阶段，解码器的$t$时刻的隐藏状态被当作query，encoder的每个时间步的hidden states作为key和value进行attention聚合. Attetion model的输出当作成上下文信息context vector，并与解码器输入$D_t$拼接起来一起送到解码器：\n\n![Image Name](https://cdn.kesci.com/upload/image/q5km7o8z93.PNG?imageView2/0/w/800/h/800)\n\n$$\nFig1具有注意机制的seq-to-seq模型解码的第二步\n$$\n\n\n下图展示了seq2seq机制的所以层的关系，下面展示了encoder和decoder的layer结构\n\n![Image Name](https://cdn.kesci.com/upload/image/q5km8dihlr.PNG?imageView2/0/w/800/h/800)\n\n$$\nFig2具有注意机制的seq-to-seq模型中层结构\n$$\n"},{"cell_type":"code","execution_count":14,"metadata":{"graffitiCellId":"id_xbbsd1x","jupyter":{},"tags":[],"slideshow":{"slide_type":"slide"},"id":"5F439000597B4D638C3C6A85BA82A82D","collapsed":false,"scrolled":false},"outputs":[],"source":"import sys\nsys.path.append('/home/kesci/input/d2len9900')\nimport d2l"},{"cell_type":"markdown","metadata":{"graffitiCellId":"id_9bq3vbt","jupyter":{},"tags":[],"slideshow":{"slide_type":"slide"},"id":"FBF33B6F0ABC45A497E1C94A67392499","mdEditEnable":false},"source":"## 解码器\n   由于带有注意机制的seq2seq的编码器与之前章节中的Seq2SeqEncoder相同，所以在此处我们只关注解码器。我们添加了一个MLP注意层(MLPAttention)，它的隐藏大小与解码器中的LSTM层相同。然后我们通过从编码器传递三个参数来初始化解码器的状态:\n   \n- the encoder outputs of all timesteps：encoder输出的各个状态，被用于attetion layer的memory部分，有相同的key和values\n\n\n- the hidden state of the encoder’s final timestep：编码器最后一个时间步的隐藏状态，被用于初始化decoder 的hidden state\n\n\n- the encoder valid length: 编码器的有效长度，借此，注意层不会考虑编码器输出中的填充标记（Paddings）\n\n\n   在解码的每个时间步，我们使用解码器的最后一个RNN层的输出作为注意层的query。然后，将注意力模型的输出与输入嵌入向量连接起来，输入到RNN层。虽然RNN层隐藏状态也包含来自解码器的历史信息，但是attention model的输出显式地选择了enc_valid_len以内的编码器输出，这样attention机制就会尽可能排除其他不相关的信息。"},{"cell_type":"code","execution_count":15,"metadata":{"graffitiCellId":"id_n3xn2jo","jupyter":{},"tags":[],"slideshow":{"slide_type":"slide"},"id":"421CF4D81A7E43378AF5C53236DC0586","collapsed":false,"scrolled":false},"outputs":[],"source":"class Seq2SeqAttentionDecoder(d2l.Decoder):\n    def __init__(self, vocab_size, embed_size, num_hiddens, num_layers,\n                 dropout=0, **kwargs):\n        super(Seq2SeqAttentionDecoder, self).__init__(**kwargs)\n        self.attention_cell = MLPAttention(num_hiddens,num_hiddens, dropout)\n        self.embedding = nn.Embedding(vocab_size, embed_size)\n        self.rnn = nn.LSTM(embed_size+ num_hiddens,num_hiddens, num_layers, dropout=dropout)\n        self.dense = nn.Linear(num_hiddens,vocab_size)\n\n    def init_state(self, enc_outputs, enc_valid_len, *args):\n        outputs, hidden_state = enc_outputs\n#         print(\"first:\",outputs.size(),hidden_state[0].size(),hidden_state[1].size())\n        # Transpose outputs to (batch_size, seq_len, hidden_size)\n        return (outputs.permute(1,0,-1), hidden_state, enc_valid_len)\n        #outputs.swapaxes(0, 1)\n        \n    def forward(self, X, state):\n        enc_outputs, hidden_state, enc_valid_len = state\n        #(\"X.size\",X.size())\n        X = self.embedding(X).transpose(0,1)\n#         print(\"Xembeding.size2\",X.size())\n        outputs = []\n        for l, x in enumerate(X):\n#             print(f\"\\n{l}-th token\")\n#             print(\"x.first.size()\",x.size())\n            # query shape: (batch_size, 1, hidden_size)\n            # select hidden state of the last rnn layer as query\n            query = hidden_state[0][-1].unsqueeze(1) # np.expand_dims(hidden_state[0][-1], axis=1)\n            # context has same shape as query\n#             print(\"query enc_outputs, enc_outputs:\\n\",query.size(), enc_outputs.size(), enc_outputs.size())\n            context = self.attention_cell(query, enc_outputs, enc_outputs, enc_valid_len)\n            # Concatenate on the feature dimension\n#             print(\"context.size:\",context.size())\n            x = torch.cat((context, x.unsqueeze(1)), dim=-1)\n            # Reshape x to (1, batch_size, embed_size+hidden_size)\n#             print(\"rnn\",x.size(), len(hidden_state))\n            out, hidden_state = self.rnn(x.transpose(0,1), hidden_state)\n            outputs.append(out)\n        outputs = self.dense(torch.cat(outputs, dim=0))\n        return outputs.transpose(0, 1), [enc_outputs, hidden_state,\n                                        enc_valid_len]"},{"cell_type":"markdown","metadata":{"graffitiCellId":"id_wbihjkp","jupyter":{},"tags":[],"slideshow":{"slide_type":"slide"},"id":"CA550D7B430D4907857B054FC17C77ED","mdEditEnable":false},"source":"现在我们可以用注意力模型来测试seq2seq。为了与第9.7节中的模型保持一致，我们对vocab_size、embed_size、num_hiddens和num_layers使用相同的超参数。结果，我们得到了相同的解码器输出形状，但是状态结构改变了。"},{"cell_type":"code","execution_count":16,"metadata":{"graffitiCellId":"id_5u9j8bu","jupyter":{},"tags":[],"slideshow":{"slide_type":"slide"},"id":"4CCF2FAC8DE741D2B86E509FFA6F33DB","collapsed":false,"scrolled":false},"outputs":[{"output_type":"stream","text":"batch size=4\nseq_length=7\nhidden dim=16\nnum_layers=2\n\nencoder output size: torch.Size([7, 4, 16])\nencoder hidden size: torch.Size([2, 4, 16])\nencoder memory size: torch.Size([2, 4, 16])\n","name":"stdout"},{"output_type":"execute_result","metadata":{},"data":{"text/plain":"(torch.Size([4, 7, 10]), 3, torch.Size([4, 7, 16]), 2, torch.Size([2, 4, 16]))"},"transient":{},"execution_count":16}],"source":"encoder = d2l.Seq2SeqEncoder(vocab_size=10, embed_size=8,\n                            num_hiddens=16, num_layers=2)\n# encoder.initialize()\ndecoder = Seq2SeqAttentionDecoder(vocab_size=10, embed_size=8,\n                                  num_hiddens=16, num_layers=2)\nX = torch.zeros((4, 7),dtype=torch.long)\nprint(\"batch size=4\\nseq_length=7\\nhidden dim=16\\nnum_layers=2\\n\")\nprint('encoder output size:', encoder(X)[0].size())\nprint('encoder hidden size:', encoder(X)[1][0].size())\nprint('encoder memory size:', encoder(X)[1][1].size())\nstate = decoder.init_state(encoder(X), None)\nout, state = decoder(X, state)\nout.shape, len(state), state[0].shape, len(state[1]), state[1][0].shape"},{"cell_type":"markdown","metadata":{"graffitiCellId":"id_k53lcnp","jupyter":{},"tags":[],"slideshow":{"slide_type":"slide"},"id":"25FB3334E1684E59966D79EA35B050D0","mdEditEnable":false},"source":"# 训练\n\n  与第9.7.4节相似，通过应用相同的训练超参数和相同的训练损失来尝试一个简单的娱乐模型。从结果中我们可以看出，由于训练数据集中的序列相对较短，额外的注意层并没有带来显著的改进。由于编码器和解码器的注意层的计算开销，该模型比没有注意的seq2seq模型慢得多。"},{"cell_type":"code","execution_count":21,"metadata":{"graffitiCellId":"id_3uv6guo","jupyter":{},"tags":[],"slideshow":{"slide_type":"slide"},"id":"6DC0F075B2EA4EA78B6D405A35D896CC","collapsed":false,"scrolled":false},"outputs":[],"source":"import zipfile\nimport torch\nimport requests\nfrom io import BytesIO\nfrom torch.utils import data\nimport sys\nimport collections\n\nclass Vocab(object): # This class is saved in d2l.\n  def __init__(self, tokens, min_freq=0, use_special_tokens=False):\n    # sort by frequency and token\n    counter = collections.Counter(tokens)\n    token_freqs = sorted(counter.items(), key=lambda x: x[0])\n    token_freqs.sort(key=lambda x: x[1], reverse=True)\n    if use_special_tokens:\n      # padding, begin of sentence, end of sentence, unknown\n      self.pad, self.bos, self.eos, self.unk = (0, 1, 2, 3)\n      tokens = ['', '', '', '']\n    else:\n      self.unk = 0\n      tokens = ['']\n    tokens += [token for token, freq in token_freqs if freq >= min_freq]\n    self.idx_to_token = []\n    self.token_to_idx = dict()\n    for token in tokens:\n      self.idx_to_token.append(token)\n      self.token_to_idx[token] = len(self.idx_to_token) - 1\n      \n  def __len__(self):\n    return len(self.idx_to_token)\n  \n  def __getitem__(self, tokens):\n    if not isinstance(tokens, (list, tuple)):\n      return self.token_to_idx.get(tokens, self.unk)\n    else:\n      return [self.__getitem__(token) for token in tokens]\n    \n  def to_tokens(self, indices):\n    if not isinstance(indices, (list, tuple)):\n      return self.idx_to_token[indices]\n    else:\n      return [self.idx_to_token[index] for index in indices]\n\ndef load_data_nmt(batch_size, max_len, num_examples=1000):\n    \"\"\"Download an NMT dataset, return its vocabulary and data iterator.\"\"\"\n    # Download and preprocess\n    def preprocess_raw(text):\n        text = text.replace('\\u202f', ' ').replace('\\xa0', ' ')\n        out = ''\n        for i, char in enumerate(text.lower()):\n            if char in (',', '!', '.') and text[i-1] != ' ':\n                out += ' '\n            out += char\n        return out \n\n\n    with open('/home/kesci/input/fraeng6506/fra.txt', 'r') as f:\n      raw_text = f.read()\n\n\n    text = preprocess_raw(raw_text)\n\n    # Tokenize\n    source, target = [], []\n    for i, line in enumerate(text.split('\\n')):\n        if i >= num_examples:\n            break\n        parts = line.split('\\t')\n        if len(parts) >= 2:\n            source.append(parts[0].split(' '))\n            target.append(parts[1].split(' '))\n\n    # Build vocab\n    def build_vocab(tokens):\n        tokens = [token for line in tokens for token in line]\n        return Vocab(tokens, min_freq=3, use_special_tokens=True)\n    src_vocab, tgt_vocab = build_vocab(source), build_vocab(target)\n\n    # Convert to index arrays\n    def pad(line, max_len, padding_token):\n        if len(line) > max_len:\n            return line[:max_len]\n        return line + [padding_token] * (max_len - len(line))\n\n    def build_array(lines, vocab, max_len, is_source):\n        lines = [vocab[line] for line in lines]\n        if not is_source:\n            lines = [[vocab.bos] + line + [vocab.eos] for line in lines]\n        array = torch.tensor([pad(line, max_len, vocab.pad) for line in lines])\n        valid_len = (array != vocab.pad).sum(1)\n        return array, valid_len\n\n    src_vocab, tgt_vocab = build_vocab(source), build_vocab(target)\n    src_array, src_valid_len = build_array(source, src_vocab, max_len, True)\n    tgt_array, tgt_valid_len = build_array(target, tgt_vocab, max_len, False)\n    train_data = data.TensorDataset(src_array, src_valid_len, tgt_array, tgt_valid_len)\n    train_iter = data.DataLoader(train_data, batch_size, shuffle=True)\n    return src_vocab, tgt_vocab, train_iter"},{"cell_type":"code","execution_count":18,"metadata":{"graffitiCellId":"id_uwt29qm","jupyter":{},"tags":[],"slideshow":{"slide_type":"slide"},"id":"16B6DBF49DE6462CB969E7FEB2B80EA4","collapsed":false,"scrolled":false},"outputs":[],"source":"embed_size, num_hiddens, num_layers, dropout = 32, 32, 2, 0.0\nbatch_size, num_steps = 64, 10\nlr, num_epochs, ctx = 0.005, 500, d2l.try_gpu()\n\nsrc_vocab, tgt_vocab, train_iter = load_data_nmt(batch_size, num_steps)\nencoder = d2l.Seq2SeqEncoder(\n    len(src_vocab), embed_size, num_hiddens, num_layers, dropout)\ndecoder = Seq2SeqAttentionDecoder(\n    len(tgt_vocab), embed_size, num_hiddens, num_layers, dropout)\nmodel = d2l.EncoderDecoder(encoder, decoder)"},{"cell_type":"markdown","metadata":{"graffitiCellId":"id_l19wr5o","jupyter":{},"tags":[],"slideshow":{"slide_type":"slide"},"id":"B087A130ADCD4C6A855201B68C1B437F","mdEditEnable":false},"source":"## 训练和预测"},{"cell_type":"code","execution_count":19,"metadata":{"graffitiCellId":"id_l50r9v2","scrolled":false,"jupyter":{},"tags":[],"slideshow":{"slide_type":"slide"},"id":"C85DB18574124B778FF37F0C939A87CF","collapsed":false},"outputs":[{"output_type":"stream","text":"epoch   50,loss 0.104, time 54.7 sec\nepoch  100,loss 0.046, time 54.8 sec\nepoch  150,loss 0.031, time 54.7 sec\nepoch  200,loss 0.027, time 54.3 sec\nepoch  250,loss 0.025, time 54.3 sec\nepoch  300,loss 0.024, time 54.4 sec\nepoch  350,loss 0.024, time 54.4 sec\nepoch  400,loss 0.024, time 54.5 sec\nepoch  450,loss 0.023, time 54.4 sec\nepoch  500,loss 0.023, time 54.7 sec\n","name":"stdout"}],"source":"d2l.train_s2s_ch9(model, train_iter, lr, num_epochs, ctx)"},{"cell_type":"code","execution_count":20,"metadata":{"graffitiCellId":"id_fgsubff","jupyter":{},"tags":[],"slideshow":{"slide_type":"slide"},"id":"59CD0BDF632440938BF388ACCD017A09","collapsed":false,"scrolled":false},"outputs":[{"output_type":"stream","text":"Go . => va !\nGood Night ! =>   !\nI'm OK . => ça va .\nI won ! => j'ai gagné !\n","name":"stdout"}],"source":"for sentence in ['Go .', 'Good Night !', \"I'm OK .\", 'I won !']:\n    print(sentence + ' => ' + d2l.predict_s2s_ch9(\n        model, sentence, src_vocab, tgt_vocab, num_steps, ctx))"}],"metadata":{"kernelspec":{"name":"python3","display_name":"Python 3","language":"python"},"language_info":{"name":"python","version":"3.7.3","mimetype":"text/x-python","codemirror_mode":{"name":"ipython","version":3},"pygments_lexer":"ipython3","nbconvert_exporter":"python","file_extension":".py"}},"nbformat":4,"nbformat_minor":1}