Files
Qwen3-4B-Chemistry-SDPO/training_score.csv
ModelHub XC 5778752181 初始化项目,由ModelHub XC社区提供模型
Model: SeongryongJung/Qwen3-4B-Chemistry-SDPO
Source: Original Platform
2026-08-13 12:49:18 +08:00

102 lines
7.6 KiB
CSV

step,critic_score_mean,critic_rewards_mean,actor_pg_loss,actor_grad_norm,actor_lr
1,0.4619140625,0.4619140625,0.28884953447050066,0.5260425396263599,0.0
2,0.40478515625,0.40478515625,0.24755145346898644,0.6269320175051689,1.0000000000000002e-06
3,0.392578125,0.392578125,0.2559864226523132,0.5038167387247086,2.0000000000000003e-06
4,0.4560546875,0.4560546875,0.2934477553862962,0.32646144926548004,3e-06
5,0.537109375,0.537109375,0.3006920863263076,0.2942509204149246,4.000000000000001e-06
6,0.56591796875,0.56591796875,0.2888475441177434,0.2903786040842533,5e-06
7,0.58544921875,0.58544921875,0.29298103353357874,0.28050167858600616,6e-06
8,0.646484375,0.646484375,0.27898271255980944,0.3003412187099457,7e-06
9,0.625,0.625,0.26837311482631776,0.33079274371266365,8.000000000000001e-06
10,0.66015625,0.66015625,0.2681529514029535,0.3171246573328972,9e-06
11,0.61669921875,0.61669921875,0.22172261147079553,0.2211923860013485,1e-05
12,0.66845703125,0.66845703125,0.20182611252062088,0.20460318587720394,1e-05
13,0.7216796875,0.7216796875,0.18659049875184053,0.19378715753555298,1e-05
14,0.7373046875,0.7373046875,0.16752206152089144,0.18391559086740017,1e-05
15,0.70361328125,0.70361328125,0.14444047730009046,0.16268958617001772,1e-05
16,0.76123046875,0.76123046875,0.12933228262198604,0.16548040322959423,1e-05
17,0.76611328125,0.76611328125,0.13829398122064251,0.18448304571211338,1e-05
18,0.744140625,0.744140625,0.12660197631654846,0.15110095590353012,1e-05
19,0.70556640625,0.70556640625,0.09941128022744561,0.14920865464955568,1e-05
20,0.77392578125,0.77392578125,0.10197432164142972,0.1752926716580987,1e-05
21,0.74853515625,0.74853515625,0.09705028147007444,0.15646957699209452,1e-05
22,0.78076171875,0.78076171875,0.07129736782177742,0.1482186522334814,1e-05
23,0.81787109375,0.81787109375,0.07421101519193884,0.15683260653167963,1e-05
24,0.85595703125,0.85595703125,0.06749791833964358,0.17304183449596167,1e-05
25,0.78857421875,0.78857421875,0.07001528055914763,0.14193351194262505,1e-05
26,0.76416015625,0.76416015625,0.0644787280053214,0.2585673490539193,1e-05
27,0.79345703125,0.79345703125,0.06803410498200013,0.1810189802199602,1e-05
28,0.7333984375,0.7333984375,0.06175067429363468,0.13102427497506142,1e-05
29,0.7783203125,0.7783203125,0.046804449997635,0.12431975454092026,1e-05
30,0.77392578125,0.77392578125,0.038438967182645944,0.1559569938108325,1e-05
31,0.76025390625,0.76025390625,0.04647161911737718,0.1456234361976385,1e-05
32,0.7939453125,0.7939453125,0.04534947794201416,0.11933708935976028,1e-05
33,0.83642578125,0.83642578125,0.04636383052692483,0.14430065639317036,1e-05
34,0.8115234375,0.8115234375,0.046606987584560056,0.1495359344407916,1e-05
35,0.76025390625,0.76025390625,0.03932185595971438,0.14291375409811735,1e-05
36,0.82177734375,0.82177734375,0.034956123329258126,0.133905871771276,1e-05
37,0.81640625,0.81640625,0.04182046812324547,0.14756610058248043,1e-05
38,0.8037109375,0.8037109375,0.039731805298444556,0.12246765289455652,1e-05
39,0.8115234375,0.8115234375,0.03831675903693271,0.11949046980589628,1e-05
40,0.8232421875,0.8232421875,0.026098412165211693,0.12175836879760027,1e-05
41,0.80810546875,0.80810546875,0.028260076234976284,0.14520351495593786,1e-05
42,0.833984375,0.833984375,0.029035456487600975,0.17687818594276905,1e-05
43,0.8193359375,0.8193359375,0.03356245695739801,0.14046786725521088,1e-05
44,0.82763671875,0.82763671875,0.03284355716304188,0.1394830783829093,1e-05
45,0.8154296875,0.8154296875,0.030147581057766826,0.11655384581536055,1e-05
46,0.8037109375,0.8037109375,0.028626308125407562,0.1188533054664731,1e-05
47,0.8212890625,0.8212890625,0.030754211201014847,0.15727455727756023,1e-05
48,0.8125,0.8125,0.01702313605133643,0.12485878681764007,1e-05
49,0.79541015625,0.79541015625,0.022401610582767573,0.1398537065833807,1e-05
50,0.7666015625,0.7666015625,0.01651571941005997,0.12312243971973658,1e-05
51,0.82421875,0.82421875,0.019167081617302795,0.1778974775224924,1e-05
52,0.83056640625,0.83056640625,0.017245586529491597,0.13062408566474915,1e-05
53,0.7958984375,0.7958984375,0.018563246627604557,0.14377865940332413,1e-05
54,0.77001953125,0.77001953125,0.027199650804331732,0.11176951322704554,1e-05
55,0.7548828125,0.7548828125,0.022619280173621026,0.10931221954524517,1e-05
56,0.78466796875,0.78466796875,0.02518373694085163,0.11203244049102068,1e-05
57,0.806640625,0.806640625,0.02106921868745626,0.11612797528505325,1e-05
58,0.8388671875,0.8388671875,0.0234936641441833,0.12233779951930046,1e-05
59,0.77734375,0.77734375,0.022023350492739735,0.12837411183863878,1e-05
60,0.82470703125,0.82470703125,0.027537235781218472,0.10643776692450047,1e-05
61,0.79931640625,0.79931640625,0.023608889111028875,0.14421263337135315,1e-05
62,0.83984375,0.83984375,0.02296517058724723,0.17886322550475597,1e-05
63,0.8369140625,0.8369140625,0.024634619662815743,0.12835557339712977,1e-05
64,0.85009765625,0.85009765625,0.016506976756064606,0.1620845478028059,1e-05
65,0.84912109375,0.84912109375,0.029049980604222536,0.14439518796280026,1e-05
66,0.82666015625,0.82666015625,0.024997733779735853,0.1297615384683013,1e-05
67,0.77978515625,0.77978515625,0.022696352618785554,0.12116180686280131,1e-05
68,0.79150390625,0.79150390625,0.018529522613441318,0.09987118374556303,1e-05
69,0.779296875,0.779296875,0.01803793113833313,0.09625563863664865,1e-05
70,0.84130859375,0.84130859375,0.01723563466374517,0.0984520148485899,1e-05
71,0.8193359375,0.8193359375,0.011416638097240972,0.0888823913410306,1e-05
72,0.83056640625,0.83056640625,0.011656536246846705,0.10679100826382637,1e-05
73,0.833984375,0.833984375,0.009777136965028688,0.11697760969400406,1e-05
74,0.8466796875,0.8466796875,0.012428767783843109,0.1301619834266603,1e-05
75,0.841796875,0.841796875,0.010418565633311838,0.08544752467423677,1e-05
76,0.79248046875,0.79248046875,0.012255287148658783,0.12487258855253458,1e-05
77,0.83642578125,0.83642578125,0.016461368026765094,0.08950501075014472,1e-05
78,0.8212890625,0.8212890625,0.016409987746085153,0.0834989370778203,1e-05
79,0.82470703125,0.82470703125,0.012333121332110675,0.09872396942228079,1e-05
80,0.84912109375,0.84912109375,0.015339945544657813,0.08171465806663036,1e-05
81,0.7919921875,0.7919921875,0.01402841219734205,0.09967908822000027,1e-05
82,0.83935546875,0.83935546875,0.011125817819977346,0.08926912536844611,1e-05
83,0.82763671875,0.82763671875,0.013762371046101396,0.1026407377794385,1e-05
84,0.83349609375,0.83349609375,0.012345594610143978,0.09586986340582371,1e-05
85,0.83447265625,0.83447265625,0.011207171866921398,0.11796947801485658,1e-05
86,0.87646484375,0.87646484375,0.015513428795949774,0.11811407143250108,1e-05
87,0.80810546875,0.80810546875,0.011633481607374235,0.08771005645394325,1e-05
88,0.80517578125,0.80517578125,0.011895685154357016,0.09411695599555969,1e-05
89,0.82958984375,0.82958984375,0.014593475944774281,0.11624695966020226,1e-05
90,0.873046875,0.873046875,0.01473958985055318,0.08287174021825194,1e-05
91,0.759765625,0.759765625,0.013241630412548389,0.09951207262929529,1e-05
92,0.8037109375,0.8037109375,0.012352060792824748,0.0821364822331816,1e-05
93,0.8251953125,0.8251953125,0.008762107602602959,0.09710145276039839,1e-05
94,0.8203125,0.8203125,0.012700424135600352,0.09056714153848588,1e-05
95,0.8251953125,0.8251953125,0.011330369384006167,0.07927328860387206,1e-05
96,0.8359375,0.8359375,0.00902163716133593,0.10526461573317647,1e-05
97,0.8212890625,0.8212890625,0.012974708422514525,0.08692156709730625,1e-05
98,0.84033203125,0.84033203125,0.014152242267626847,0.09331048559397459,1e-05
99,0.8193359375,0.8193359375,0.01067210086236317,0.07151691475883126,1e-05
100,0.84521484375,0.84521484375,0.007015153896071297,0.1468870275421068,1e-05