过程
/usr/bin/python3.5 /home/zqy/lunwen/word_vec.py
Found and verified text8.zip
Data size 17005207
Most common words (+UNK) [['UNK', 418391], ('the', 1061396), ('of', 593677), ('and', 416629), ('one', 411764)]
Sample data [5237, 3082, 12, 6, 195, 2, 3137, 46, 59, 156] ['anarchism', 'originated', 'as', 'a', 'term', 'of', 'abuse', 'first', 'used', 'against']
3082 originated -> 12 as
3082 originated -> 5237 anarchism
12 as -> 3082 originated
12 as -> 6 a
6 a -> 195 term
6 a -> 12 as
195 term -> 6 a
195 term -> 2 of
2018-01-13 22:47:41.860462: W tensorflow/core/platform/cpu_feature_guard.cc:45] The TensorFlow library wasn't compiled to use SSE4.1 instructions, but these are available on your machine and could speed up CPU computations.
2018-01-13 22:47:41.860552: W tensorflow/core/platform/cpu_feature_guard.cc:45] The TensorFlow library wasn't compiled to use SSE4.2 instructions, but these are available on your machine and could speed up CPU computations.
2018-01-13 22:47:41.860567: W tensorflow/core/platform/cpu_feature_guard.cc:45] The TensorFlow library wasn't compiled to use AVX instructions, but these are available on your machine and could speed up CPU computations.
Initialized
Average loss at step 0 : 259.220703125
Nearest to b: depreciation, dynasty, opportunity, cabins, firewood, great, feathered, achish,
Nearest to would: bertrand, uncontroversial, milking, salk, ph, beecher, designers, brod,
Nearest to over: gonzo, redstone, universidade, positioning, logarithm, weasels, glut, wavering,
Nearest to new: byline, norfolk, esso, proactive, slip, mexican, koalas, metaphorical,
Nearest to than: tauris, conspiratorial, vor, matteo, hurts, mantle, rosalyn, patents,
Nearest to UNK: union, simpson, diy, mantle, opening, possess, reviled, troubadors,
Nearest to nine: bilateral, cdr, cedar, ffg, citizens, proviso, shaping, crusades,
Nearest to five: saltire, communist, stilicho, motets, recreational, evaluate, sheaf, tennant,
Nearest to may: prickly, alfaro, mooring, hekate, koichi, loosened, umbrella, imagines,
Nearest to is: whatsoever, compact, stationary, wane, batch, periphery, duran, fasci,
Nearest to eight: rutskoy, laureate, disclosures, occasional, fetching, severly, galton, puppeteer,
Nearest to who: criminality, aspect, emeritus, moshe, extradition, caving, jourdan, catapulted,
Nearest to state: perceptive, riviera, psoriasis, sherry, pharaonic, endorsement, pollack, anacreon,
Nearest to such: minicomputers, colliding, piggy, creativity, liars, yen, pelagius, klein,
Nearest to as: cardboard, lecture, oliphant, schilling, mars, phillip, chogm, poaching,
Nearest to seven: divers, oxygen, gurdjieff, reformatory, frivolity, freiherr, gabriele, downstream,
Average loss at step 2000 : 114.259895983
Average loss at step 4000 : 52.7314318206
Average loss at step 6000 : 33.4762375052
Average loss at step 8000 : 23.1981662744
Average loss at step 10000 : 17.9444110692
Nearest to b: lymphoma, opportunity, organs, dynasty, anatomy, great, yields, chadic,
Nearest to would: bertrand, ph, buried, attendance, gb, ability, chain, detachment,
Nearest to over: burned, mis, legs, referred, nine, dog, mainland, potato,
Nearest to new: levant, mexican, spirited, sacks, north, metaphorical, essence, august,
Nearest to than: maine, holidays, var, canaris, operated, gui, grains, hand,
Nearest to UNK: archie, and, vs, coke, one, lymphoma, the, a,
Nearest to nine: zero, archie, lymphoma, one, four, eight, six, gland,
Nearest to five: basins, nine, one, eight, zero, six, jpg, four,
Nearest to may: restricted, rebuild, coast, heights, nearly, nine, disorders, transport,
Nearest to is: was, are, and, nine, forall, has, called, uruguayan,
Nearest to eight: zero, nine, one, archie, five, bckgr, three, lymphoma,
Nearest to who: criminality, principle, alliance, grain, agave, and, arcade, cornwallis,
Nearest to state: lymphoma, bckgr, catch, endorsement, pharaonic, circumstances, tubing, chronicles,
Nearest to such: well, creativity, a, basins, inspired, ki, vessels, vs,
Nearest to as: in, operation, player, lecture, lymphoma, by, abba, during,
Nearest to seven: nine, zero, two, lymphoma, jpg, archie, five, one,
Average loss at step 12000 : 13.8751676543
Average loss at step 14000 : 11.7038348273
Average loss at step 16000 : 9.99806536496
Average loss at step 18000 : 8.55141812801
Average loss at step 20000 : 7.92424252069
Nearest to b: d, lymphoma, and, yields, organs, opportunity, circ, great,
Nearest to would: bertrand, ph, to, agouti, attendance, is, detachment, chain,
Nearest to over: and, as, mis, burned, legs, five, on, hbox,
Nearest to new: levant, spirited, north, high, metaphorical, dasyprocta, essence, ask,
Nearest to than: maine, or, holidays, exercises, and, operated, patents, gui,
Nearest to UNK: dasyprocta, archie, agouti, circ, and, operatorname, trapezohedron, one,
Nearest to nine: eight, six, seven, zero, five, four, three, agouti,
Nearest to five: eight, nine, four, six, zero, two, seven, dasyprocta,
Nearest to may: moor, can, restricted, nine, rebuild, heights, cooler, antiprism,
Nearest to is: was, are, has, nine, were, circ, in, forall,
Nearest to eight: nine, six, zero, five, four, seven, three, two,
Nearest to who: and, which, criminality, it, agouti, but, threshold, agave,
Nearest to state: lymphoma, chronicles, pharaonic, bckgr, endorsement, hecuba, dasyprocta, catch,
Nearest to such: pelagius, well, circ, creativity, vessels, a, dasyprocta, yen,
Nearest to as: circ, and, dasyprocta, for, agouti, by, operation, was,
Nearest to seven: nine, eight, four, six, zero, five, two, three,
Average loss at step 22000 : 6.92497203529
Average loss at step 24000 : 6.82812125087
Average loss at step 26000 : 6.63320548046
Average loss at step 28000 : 6.41591437364
Average loss at step 30000 : 5.95010187364
Nearest to b: d, and, lymphoma, accusations, yields, organs, gatherers, great,
Nearest to would: bertrand, to, ph, can, could, may, agouti, indexes,
Nearest to over: on, mis, as, burned, positioning, and, cancerous, five,
Nearest to new: levant, high, essence, north, spirited, metaphorical, dasyprocta, ask,
Nearest to than: or, maine, dhamma, exercises, holidays, gui, patents, operated,
Nearest to UNK: dasyprocta, archie, agouti, circ, two, operatorname, five, trapezohedron,
Nearest to nine: eight, six, seven, five, four, zero, three, agouti,
Nearest to five: four, six, eight, seven, zero, nine, three, two,
Nearest to may: can, should, moor, would, rebuild, nine, heights, could,
Nearest to is: was, are, has, were, circ, be, forall, as,
Nearest to eight: nine, six, four, seven, five, zero, three, circ,
Nearest to who: and, which, he, it, they, agouti, threshold, criminality,
Nearest to state: lymphoma, perceptive, chronicles, pharaonic, appendix, endorsement, bckgr, hecuba,
Nearest to such: pelagius, well, circ, yen, creativity, leisure, vessels, dasyprocta,
Nearest to as: circ, agouti, dasyprocta, by, lymphoma, operation, operatorname, with,
Nearest to seven: eight, nine, five, six, four, zero, three, two,
Average loss at step 32000 : 5.98741295695
Average loss at step 34000 : 5.74154632425
Average loss at step 36000 : 5.75207448113
Average loss at step 38000 : 5.52982545102
Average loss at step 40000 : 5.22882912374
Nearest to b: d, UNK, lymphoma, circ, eight, yields, accusations, opportunity,
Nearest to would: bertrand, can, could, to, may, will, ph, must,
Nearest to over: mis, on, cancerous, akita, and, burned, positioning, five,
Nearest to new: levant, high, metaphorical, spirited, essence, dasyprocta, north, catch,
Nearest to than: or, maine, dhamma, exercises, holidays, gui, patents, billion,
Nearest to UNK: dasyprocta, archie, agouti, circ, four, operatorname, three, recitative,
Nearest to nine: eight, seven, six, zero, five, four, three, agouti,
Nearest to five: four, eight, six, seven, three, zero, two, nine,
Nearest to may: can, should, would, will, moor, could, heights, rebuild,
Nearest to is: was, are, has, were, forall, became, while, circ,
Nearest to eight: seven, six, four, nine, five, zero, three, two,
Nearest to who: which, he, and, they, it, reformist, agouti, specifications,
Nearest to state: lymphoma, perceptive, pharaonic, chronicles, dasyprocta, endorsement, agouti, appendix,
Nearest to such: well, pelagius, metroid, circ, continent, leisure, yen, many,
Nearest to as: circ, agouti, dasyprocta, lymphoma, by, operatorname, in, with,
Nearest to seven: eight, six, four, five, nine, zero, three, two,
Average loss at step 42000 : 5.384367715
Average loss at step 44000 : 5.24350860226
Average loss at step 46000 : 5.21256376803
Average loss at step 48000 : 5.24430999529
Average loss at step 50000 : 4.98843484068
Nearest to b: d, kapoor, lymphoma, UNK, circ, f, six, eight,
Nearest to would: can, will, could, bertrand, to, may, must, might,
Nearest to over: on, kapoor, cancerous, mis, and, four, seven, three,
Nearest to new: levant, kapoor, spirited, high, essence, metaphorical, dasyprocta, analyst,
Nearest to than: or, maine, and, dhamma, for, exercises, kapoor, billion,
Nearest to UNK: kapoor, dasyprocta, agouti, archie, operatorname, circ, five, recitative,
Nearest to nine: eight, six, seven, five, zero, three, four, kapoor,
Nearest to five: six, four, three, eight, seven, two, zero, dasyprocta,
Nearest to may: can, should, would, will, could, moor, kapoor, three,
Nearest to is: was, are, has, were, be, kapoor, forall, while,
Nearest to eight: six, seven, nine, four, five, three, zero, kapoor,
Nearest to who: which, he, and, they, it, mangeshkar, specifications, reformist,
Nearest to state: lymphoma, pharaonic, perceptive, chronicles, pasteur, appendix, endorsement, cnn,
Nearest to such: well, pelagius, witty, tragic, many, known, circ, continent,
Nearest to as: agouti, circ, kapoor, dasyprocta, mukherjee, lymphoma, operatorname, and,
Nearest to seven: eight, four, six, five, three, nine, zero, two,
Average loss at step 52000 : 5.02020378375
Average loss at step 54000 : 5.17726987147
Average loss at step 56000 : 5.03323707342
Average loss at step 58000 : 5.0789374454
Average loss at step 60000 : 4.94255949724
Nearest to b: d, UNK, f, kapoor, yields, lymphoma, circ, limbs,
Nearest to would: can, will, could, may, might, must, bertrand, to,
Nearest to over: on, five, logarithm, kapoor, ursus, cancerous, positioning, akita,
Nearest to new: levant, kapoor, spirited, pulau, michelob, essence, dasyprocta, analyst,
Nearest to than: or, wct, maine, dhamma, billion, kapoor, for, exercises,
Nearest to UNK: kapoor, ursus, dasyprocta, archie, agouti, michelob, operatorname, five,
Nearest to nine: eight, six, seven, four, five, ursus, zero, kapoor,
Nearest to five: six, four, three, eight, seven, nine, dasyprocta, two,
Nearest to may: can, should, would, could, will, kapoor, moor, to,
Nearest to is: was, are, has, ursus, kapoor, mukherjee, wct, michelob,
Nearest to eight: six, nine, seven, four, five, three, zero, kapoor,
Nearest to who: he, which, they, ursus, it, specifications, mangeshkar, never,
Nearest to state: lymphoma, pharaonic, chronicles, perceptive, pasteur, aveiro, riviera, agouti,
Nearest to such: well, pelagius, known, many, tragic, minicomputers, continent, leisure,
Nearest to as: agouti, circ, kapoor, ursus, dasyprocta, in, mukherjee, lymphoma,
Nearest to seven: eight, six, five, nine, four, three, zero, kapoor,
Average loss at step 62000 : 5.0228234508
Average loss at step 64000 : 4.82479051709
Average loss at step 66000 : 4.60096224284
Average loss at step 68000 : 4.96966059649
Average loss at step 70000 : 4.90493321407
Nearest to b: d, UNK, f, seven, kapoor, thaler, circ, c,
Nearest to would: can, will, could, may, might, must, should, bertrand,
Nearest to over: on, microcebus, logarithm, mico, positioning, cebus, four, akita,
Nearest to new: levant, kapoor, spirited, michelob, pulau, essence, analyst, dasyprocta,
Nearest to than: or, wct, maine, and, billion, dhamma, but, geared,
Nearest to UNK: kapoor, dasyprocta, cebus, ursus, archie, agouti, circ, mico,
Nearest to nine: eight, seven, six, five, four, three, zero, ursus,
Nearest to five: six, three, four, eight, seven, nine, two, zero,
Nearest to may: can, would, should, will, could, cebus, must, might,
Nearest to is: was, are, has, ursus, cebus, wct, while, kapoor,
Nearest to eight: six, nine, seven, five, four, three, zero, kapoor,
Nearest to who: he, which, they, and, specifications, reformist, never, it,
Nearest to state: lymphoma, thaler, chronicles, perceptive, aveiro, riviera, pharaonic, agouti,
Nearest to such: well, pelagius, known, many, these, minicomputers, few, tragic,
Nearest to as: circ, agouti, kapoor, ursus, dasyprocta, mukherjee, lymphoma, by,
Nearest to seven: eight, six, five, nine, four, three, zero, ursus,
Average loss at step 72000 : 4.7390078398
Average loss at step 74000 : 4.82284517503
Average loss at step 76000 : 4.72336018646
Average loss at step 78000 : 4.79799332559
Average loss at step 80000 : 4.78921988702
Nearest to b: d, UNK, f, thaler, kapoor, circ, c, lymphoma,
Nearest to would: can, may, will, could, might, must, should, to,
Nearest to over: microcebus, logarithm, mico, cebus, ursus, akita, kapoor, as,
Nearest to new: levant, kapoor, spirited, pulau, essence, michelob, dasyprocta, analyst,
Nearest to than: or, and, wct, maine, but, dhamma, billion, matteo,
Nearest to UNK: kapoor, ursus, archie, agouti, dasyprocta, cebus, iit, circ,
Nearest to nine: eight, six, seven, five, four, zero, ursus, kapoor,
Nearest to five: six, four, eight, seven, three, nine, zero, two,
Nearest to may: can, would, should, will, could, might, must, cebus,
Nearest to is: was, are, escuela, has, ursus, cebus, kapoor, michelob,
Nearest to eight: six, nine, seven, five, four, zero, three, kapoor,
Nearest to who: he, which, they, specifications, never, reformist, ursus, it,
Nearest to state: lymphoma, perceptive, thaler, aveiro, chronicles, agouti, pharaonic, ducas,
Nearest to such: well, pelagius, these, many, known, minicomputers, few, ethics,
Nearest to as: circ, agouti, ursus, kapoor, dasyprocta, mukherjee, lymphoma, michelob,
Nearest to seven: six, eight, five, four, nine, three, zero, one,
Average loss at step 82000 : 4.75267288053
Average loss at step 84000 : 4.73391491497
Average loss at step 86000 : 4.78696395206
Average loss at step 88000 : 4.73892472947
Average loss at step 90000 : 4.72603131425
Nearest to b: d, c, UNK, f, j, thaler, circ, kapoor,
Nearest to would: can, will, may, could, might, must, should, to,
Nearest to over: microcebus, about, logarithm, on, absorption, mico, cebus, into,
Nearest to new: levant, kapoor, circ, pulau, essence, dasyprocta, spirited, michelob,
Nearest to than: or, but, wct, dhamma, maine, billion, umayyad, and,
Nearest to UNK: ursus, kapoor, dasyprocta, cebus, agouti, archie, iit, operatorname,
Nearest to nine: eight, seven, six, five, four, zero, three, ursus,
Nearest to five: four, eight, seven, six, three, zero, nine, two,
Nearest to may: can, would, should, will, could, might, must, cannot,
Nearest to is: was, has, are, escuela, ursus, cebus, kapoor, circ,
Nearest to eight: seven, six, five, nine, four, three, zero, kapoor,
Nearest to who: he, they, which, specifications, never, and, it, ursus,
Nearest to state: perceptive, lymphoma, thaler, aveiro, pharaonic, chronicles, riviera, agouti,
Nearest to such: well, pelagius, these, many, known, some, including, few,
Nearest to as: circ, agouti, ursus, kapoor, dasyprocta, mukherjee, lymphoma, when,
Nearest to seven: eight, six, five, four, nine, three, zero, one,
Average loss at step 92000 : 4.65910807931
Average loss at step 94000 : 4.72688456666
Average loss at step 96000 : 4.68691073239
Average loss at step 98000 : 4.59641505724
Average loss at step 100000 : 4.69953029454
Nearest to b: d, c, f, circ, j, kapoor, UNK, thaler,
Nearest to would: will, can, may, could, might, must, should, to,
Nearest to over: microcebus, about, logarithm, absorption, mico, cebus, on, akita,
Nearest to new: levant, essence, kapoor, analyst, mexican, spirited, separate, pulau,
Nearest to than: or, but, billion, wct, dhamma, maine, emblematic, umayyad,
Nearest to UNK: kapoor, cebus, ursus, dasyprocta, agouti, michelob, mico, circ,
Nearest to nine: eight, seven, six, five, zero, four, three, ursus,
Nearest to five: four, seven, three, eight, six, two, nine, zero,
Nearest to may: can, would, should, will, could, might, must, cannot,
Nearest to is: was, are, has, escuela, ursus, cebus, circ, while,
Nearest to eight: seven, six, nine, five, four, zero, three, kapoor,
Nearest to who: he, they, which, and, specifications, never, litigants, also,
Nearest to state: perceptive, lymphoma, aveiro, thaler, agouti, chronicles, pharaonic, marr,
Nearest to such: well, these, pelagius, many, known, including, some, few,
Nearest to as: circ, agouti, kapoor, ursus, mukherjee, dasyprocta, constituci, lymphoma,
Nearest to seven: six, eight, five, four, nine, three, zero, circ,
结果
