Training in progress, step 100, checkpoint

last-checkpoint/global_step100/bf16_zero_pp_rank_0_mp_rank_00_optim_states.pt

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:185321aac63a9ef69c1a3bb618c9fe0ed96266c2772d61f8cc5cdbf15717c36d
+size 12470656801

last-checkpoint/global_step100/bf16_zero_pp_rank_1_mp_rank_00_optim_states.pt

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:ad9544c22be8f7e6a654977c975bed14b186d60a06908df929ab6b245e738d97
+size 12470656801

last-checkpoint/global_step100/bf16_zero_pp_rank_2_mp_rank_00_optim_states.pt

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:375ef4e8da96a0b3d79b883add51aaf0d4ee0b11fd4d209d9f4e9e147c047a72
+size 12470656801

last-checkpoint/global_step100/bf16_zero_pp_rank_3_mp_rank_00_optim_states.pt

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:adaf5ef950808ab6861031f692982176d72d6f95c8cce893163e9724a1ce25a5
+size 12470656801

last-checkpoint/global_step100/bf16_zero_pp_rank_4_mp_rank_00_optim_states.pt

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:76dc1e9234eee497f342ca9232d56499b2d279f9c91dca7f33f7c26b8388a050
+size 12470656801

last-checkpoint/global_step100/bf16_zero_pp_rank_5_mp_rank_00_optim_states.pt

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:c6221e7c45318512bcabe10b3c063bdf639465f60a0bbe51bf948dc986c9c7d4
+size 12470656801

last-checkpoint/global_step100/bf16_zero_pp_rank_6_mp_rank_00_optim_states.pt

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:c69e94850d2ee2d76c1b1a371995edfa34c12e97a54154a4cde49c2c118ff311
+size 12470656801

last-checkpoint/global_step100/bf16_zero_pp_rank_7_mp_rank_00_optim_states.pt

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:b024a3c0e4cf4af23a13e00e3168570c51da46a63d9b3874549702f86a91d280
+size 12470656801

last-checkpoint/global_step100/zero_pp_rank_0_mp_rank_00_model_states.pt

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:48ae2b22525a92a282ebfd9ffd5fd45d1f007508d2536902a3aa415c13642bbb
+size 113836509

last-checkpoint/global_step100/zero_pp_rank_1_mp_rank_00_model_states.pt

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:b6d8da73ac1c78733895b4eab36a98f3ffcf75bcbdccb4c9bcf8b36e4c104078
+size 113836509

last-checkpoint/global_step100/zero_pp_rank_2_mp_rank_00_model_states.pt

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:60ffbfffd2ee9921e5894a73d22b338e4d08adc2c436341ad5190c375ec6e57c
+size 113836509

last-checkpoint/global_step100/zero_pp_rank_3_mp_rank_00_model_states.pt

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:8d8c49d10b68e607a9f354930e677c8f62c09a1ac4047850941c99e926beb5aa
+size 113836509

last-checkpoint/global_step100/zero_pp_rank_4_mp_rank_00_model_states.pt

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:fce5fceb7a2be926e8d2fa6d3c699a78f0b44194fcd0c9469b7f1ce7351fd3ba
+size 113836509

last-checkpoint/global_step100/zero_pp_rank_5_mp_rank_00_model_states.pt

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:95bfad5c30cd6b22e67af461fba57aeb78861f9e74b2e4c524c476ece3c785ef
+size 113836509

last-checkpoint/global_step100/zero_pp_rank_6_mp_rank_00_model_states.pt

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:d5fca634c2f3ebf6ea94359b637b864661f513f987cb481f32a4f0fd4ed8610e
+size 113836509

last-checkpoint/global_step100/zero_pp_rank_7_mp_rank_00_model_states.pt

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:3a1e5f8b0f8c5e69f505ffd6ce18ad7df3adc165bb76f5e5b4467cfbc21a6f18
+size 113836509

last-checkpoint/latest

@@ -1 +1 @@
-global_step50
+global_step100

last-checkpoint/model-00001-of-00004.safetensors

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:18c16aeaace0dc1d447ca9d0194b7600e6e4b69053c80165c422ec024dde78fb
+oid sha256:4bed4480dc618d5c958149a905b27159769a03f3f48747cbc85df65ec875af59
 size 4998056552

last-checkpoint/model-00002-of-00004.safetensors

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:10b724aba4d0dfa1b78b76c3fec8cfcc384d2d44200ddf6e7c2653a35c02ce29
+oid sha256:d11e702bb5f8eca6805ddbe9ee14453556e2faf2567d389f6951139145dfe8ba
 size 4915962464

last-checkpoint/model-00003-of-00004.safetensors

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:1c0c6ac4eaf53ce24ed8cfc98c6b5fc719c9a222371f7cf30788abf69a68a820
+oid sha256:1e142472c59526c67cd73513fe90b3fad8c125caa8493c1f9f145a50bb229869
 size 4915962496

last-checkpoint/model-00004-of-00004.safetensors

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:d24b28d0d968b8e7ab6e2cb51cdcdf7f5ba8a5edc51ef7110d44cbfc7466781d
+oid sha256:b7483638aacca60c77f95421085f8cb44e640c8410d5d8b2c5e8b4a18fcd05f9
 size 2704357976

last-checkpoint/scheduler.pt

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:59965da6dfd8da04a42b4cb33b04635c1f0c18a11fcbb5554b89fc2178244bf5
+oid sha256:f522ff7240667fded1faf491e50cb0e354ba2050c394a86f93f2e93a583e166b
 size 1465

last-checkpoint/trainer_state.json

@@ -2,362 +2,712 @@
   "best_global_step": null,
   "best_metric": null,
   "best_model_checkpoint": null,
-  "epoch": 0.0354735721887194,
+  "epoch": 0.0709471443774388,
   "eval_steps": 500,
-  "global_step": 50,
+  "global_step": 100,
   "is_hyper_param_search": false,
   "is_local_process_zero": true,
   "is_world_process_zero": true,
   "log_history": [
     {
       "epoch": 0.000709471443774388,
-      "grad_norm": 265.8623418896737,
+      "grad_norm": 265.86757613893764,
       "learning_rate": 0.0,
       "loss": 2.1209,
       "step": 1
     },
     {
       "epoch": 0.001418942887548776,
-      "grad_norm": 256.940555533612,
+      "grad_norm": 256.9364781809131,
       "learning_rate": 2.1276595744680852e-07,
       "loss": 1.9508,
       "step": 2
     },
     {
       "epoch": 0.002128414331323164,
-      "grad_norm": 261.8330553257261,
+      "grad_norm": 261.72185657120684,
       "learning_rate": 4.2553191489361704e-07,
-      "loss": 2.2064,
+      "loss": 2.1995,
       "step": 3
     },
     {
       "epoch": 0.002837885775097552,
-      "grad_norm": 259.70078467704815,
+      "grad_norm": 261.0961921742672,
       "learning_rate": 6.382978723404255e-07,
-      "loss": 2.003,
+      "loss": 2.0081,
       "step": 4
     },
     {
       "epoch": 0.0035473572188719402,
-      "grad_norm": 258.9967866863972,
+      "grad_norm": 255.1019965019492,
       "learning_rate": 8.510638297872341e-07,
-      "loss": 2.1071,
+      "loss": 2.0828,
       "step": 5
     },
     {
       "epoch": 0.004256828662646328,
-      "grad_norm": 202.62138283306564,
+      "grad_norm": 201.9040896756704,
       "learning_rate": 1.0638297872340427e-06,
-      "loss": 1.6393,
+      "loss": 1.6403,
       "step": 6
     },
     {
       "epoch": 0.004966300106420717,
-      "grad_norm": 197.20410103629487,
+      "grad_norm": 198.32057240838185,
       "learning_rate": 1.276595744680851e-06,
-      "loss": 1.6007,
+      "loss": 1.6037,
       "step": 7
     },
     {
       "epoch": 0.005675771550195104,
-      "grad_norm": 91.87927102775392,
+      "grad_norm": 91.64675018232528,
       "learning_rate": 1.4893617021276596e-06,
-      "loss": 1.0148,
+      "loss": 1.0155,
       "step": 8
     },
     {
       "epoch": 0.006385242993969493,
-      "grad_norm": 67.54197329306645,
+      "grad_norm": 71.73605818031255,
       "learning_rate": 1.7021276595744682e-06,
-      "loss": 0.8182,
+      "loss": 0.8153,
       "step": 9
     },
     {
       "epoch": 0.0070947144377438804,
-      "grad_norm": 24.875987520184307,
+      "grad_norm": 24.941283646351614,
       "learning_rate": 1.9148936170212763e-06,
-      "loss": 0.3787,
+      "loss": 0.3811,
       "step": 10
     },
     {
       "epoch": 0.007804185881518269,
-      "grad_norm": 31.037715727866885,
+      "grad_norm": 31.071639133351898,
       "learning_rate": 2.1276595744680853e-06,
-      "loss": 0.3839,
+      "loss": 0.3851,
       "step": 11
     },
     {
       "epoch": 0.008513657325292657,
-      "grad_norm": 20.53729134657492,
+      "grad_norm": 20.77631646223411,
       "learning_rate": 2.340425531914894e-06,
-      "loss": 0.2913,
+      "loss": 0.2891,
       "step": 12
     },
     {
       "epoch": 0.009223128769067045,
-      "grad_norm": 13.556641404175776,
+      "grad_norm": 13.477671793796729,
       "learning_rate": 2.553191489361702e-06,
-      "loss": 0.255,
+      "loss": 0.2528,
       "step": 13
     },
     {
       "epoch": 0.009932600212841433,
-      "grad_norm": 2.5620771569256253,
+      "grad_norm": 2.6555809333853024,
       "learning_rate": 2.7659574468085106e-06,
-      "loss": 0.1531,
+      "loss": 0.1542,
       "step": 14
     },
     {
       "epoch": 0.010642071656615822,
-      "grad_norm": 2.6919652420732567,
+      "grad_norm": 2.898054397668401,
       "learning_rate": 2.978723404255319e-06,
-      "loss": 0.1764,
+      "loss": 0.1762,
       "step": 15
     },
     {
       "epoch": 0.011351543100390209,
-      "grad_norm": 2.3096891218856115,
+      "grad_norm": 2.1504034279865385,
       "learning_rate": 3.1914893617021277e-06,
-      "loss": 0.1404,
+      "loss": 0.1403,
       "step": 16
     },
     {
       "epoch": 0.012061014544164597,
-      "grad_norm": 5.009561741431305,
+      "grad_norm": 5.259173556940105,
       "learning_rate": 3.4042553191489363e-06,
-      "loss": 0.1568,
+      "loss": 0.1561,
       "step": 17
     },
     {
       "epoch": 0.012770485987938986,
-      "grad_norm": 3.9453885969385674,
+      "grad_norm": 4.169332828012237,
       "learning_rate": 3.6170212765957445e-06,
-      "loss": 0.1627,
+      "loss": 0.1632,
       "step": 18
     },
     {
       "epoch": 0.013479957431713374,
-      "grad_norm": 2.0110788572934637,
+      "grad_norm": 1.9000129489744462,
       "learning_rate": 3.829787234042553e-06,
-      "loss": 0.1692,
+      "loss": 0.1667,
       "step": 19
     },
     {
       "epoch": 0.014189428875487761,
-      "grad_norm": 7.699519076180619,
+      "grad_norm": 7.769485079957688,
       "learning_rate": 4.042553191489362e-06,
-      "loss": 0.1887,
+      "loss": 0.1893,
       "step": 20
     },
     {
       "epoch": 0.01489890031926215,
-      "grad_norm": 10.98920176026635,
+      "grad_norm": 11.504242044440007,
       "learning_rate": 4.255319148936171e-06,
-      "loss": 0.2045,
+      "loss": 0.2085,
       "step": 21
     },
     {
       "epoch": 0.015608371763036538,
-      "grad_norm": 6.341356238078373,
+      "grad_norm": 7.145142567820805,
       "learning_rate": 4.468085106382979e-06,
-      "loss": 0.1875,
+      "loss": 0.1913,
       "step": 22
     },
     {
       "epoch": 0.016317843206810925,
-      "grad_norm": 3.298166086020585,
+      "grad_norm": 2.736186486725912,
       "learning_rate": 4.680851063829788e-06,
-      "loss": 0.1742,
+      "loss": 0.1741,
       "step": 23
     },
     {
       "epoch": 0.017027314650585313,
-      "grad_norm": 6.024563422742684,
+      "grad_norm": 5.7916893585600695,
       "learning_rate": 4.893617021276596e-06,
-      "loss": 0.1418,
+      "loss": 0.1408,
       "step": 24
     },
     {
       "epoch": 0.0177367860943597,
-      "grad_norm": 4.160451566833143,
+      "grad_norm": 4.68647394202743,
       "learning_rate": 5.106382978723404e-06,
-      "loss": 0.1617,
+      "loss": 0.1632,
       "step": 25
     },
     {
       "epoch": 0.01844625753813409,
-      "grad_norm": 3.11898557000962,
+      "grad_norm": 2.0524513403445757,
       "learning_rate": 5.319148936170213e-06,
-      "loss": 0.1591,
+      "loss": 0.1563,
       "step": 26
     },
     {
       "epoch": 0.01915572898190848,
-      "grad_norm": 2.363362171092533,
+      "grad_norm": 1.588252902269814,
       "learning_rate": 5.531914893617021e-06,
-      "loss": 0.1594,
+      "loss": 0.1582,
       "step": 27
     },
     {
       "epoch": 0.019865200425682867,
-      "grad_norm": 3.432509859447859,
+      "grad_norm": 3.5755825255555225,
       "learning_rate": 5.74468085106383e-06,
-      "loss": 0.1455,
+      "loss": 0.1451,
       "step": 28
     },
     {
       "epoch": 0.020574671869457255,
-      "grad_norm": 6.769431780529344,
+      "grad_norm": 7.715437458161659,
       "learning_rate": 5.957446808510638e-06,
-      "loss": 0.1564,
+      "loss": 0.157,
       "step": 29
     },
     {
       "epoch": 0.021284143313231644,
-      "grad_norm": 4.155888936284697,
+      "grad_norm": 19.70487906339096,
       "learning_rate": 6.1702127659574465e-06,
-      "loss": 0.1659,
+      "loss": 0.1725,
       "step": 30
     },
     {
       "epoch": 0.02199361475700603,
-      "grad_norm": 4.482844757668334,
+      "grad_norm": 2.21235512385039,
       "learning_rate": 6.3829787234042555e-06,
-      "loss": 0.1387,
+      "loss": 0.1318,
       "step": 31
     },
     {
       "epoch": 0.022703086200780417,
-      "grad_norm": 8.154233295690652,
+      "grad_norm": 5.021653379215672,
       "learning_rate": 6.5957446808510645e-06,
-      "loss": 0.19,
+      "loss": 0.1795,
       "step": 32
     },
     {
       "epoch": 0.023412557644554806,
-      "grad_norm": 1.9560161370302223,
+      "grad_norm": 25.495061679606632,
       "learning_rate": 6.808510638297873e-06,
-      "loss": 0.1516,
+      "loss": 0.1586,
       "step": 33
     },
     {
       "epoch": 0.024122029088329194,
-      "grad_norm": 4.482796200223318,
+      "grad_norm": 4.7481186730757035,
       "learning_rate": 7.021276595744681e-06,
-      "loss": 0.1468,
+      "loss": 0.1552,
       "step": 34
     },
     {
       "epoch": 0.024831500532103583,
-      "grad_norm": 1.3229503448711997,
+      "grad_norm": 3.6399107060391365,
       "learning_rate": 7.234042553191489e-06,
-      "loss": 0.1417,
+      "loss": 0.1573,
       "step": 35
     },
     {
       "epoch": 0.02554097197587797,
-      "grad_norm": 1.1937407062234258,
+      "grad_norm": 3.3754400837055707,
       "learning_rate": 7.446808510638298e-06,
-      "loss": 0.1362,
+      "loss": 0.149,
       "step": 36
     },
     {
       "epoch": 0.02625044341965236,
-      "grad_norm": 2.716831485809823,
+      "grad_norm": 2.8787349588257842,
       "learning_rate": 7.659574468085105e-06,
-      "loss": 0.1438,
+      "loss": 0.154,
       "step": 37
     },
     {
       "epoch": 0.026959914863426748,
-      "grad_norm": 2.12698604859137,
+      "grad_norm": 2.7404452323712416,
       "learning_rate": 7.872340425531914e-06,
-      "loss": 0.1502,
+      "loss": 0.1563,
       "step": 38
     },
     {
       "epoch": 0.027669386307201137,
-      "grad_norm": 1.8750032390372138,
+      "grad_norm": 1.8896448248412065,
       "learning_rate": 8.085106382978723e-06,
-      "loss": 0.1621,
+      "loss": 0.1568,
       "step": 39
     },
     {
       "epoch": 0.028378857750975522,
-      "grad_norm": 4.170717575926779,
+      "grad_norm": 6.6357403507533395,
       "learning_rate": 8.297872340425532e-06,
-      "loss": 0.1553,
+      "loss": 0.1608,
       "step": 40
     },
     {
       "epoch": 0.02908832919474991,
-      "grad_norm": 2.1480680867111333,
+      "grad_norm": 5.2141592606599545,
       "learning_rate": 8.510638297872341e-06,
-      "loss": 0.1422,
+      "loss": 0.1495,
       "step": 41
     },
     {
       "epoch": 0.0297978006385243,
-      "grad_norm": 4.872070432967606,
+      "grad_norm": 2.653621690332451,
       "learning_rate": 8.723404255319149e-06,
-      "loss": 0.1618,
+      "loss": 0.1746,
       "step": 42
     },
     {
       "epoch": 0.030507272082298687,
-      "grad_norm": 5.281603269228914,
+      "grad_norm": 2.1695013496806483,
       "learning_rate": 8.936170212765958e-06,
-      "loss": 0.1563,
+      "loss": 0.1501,
       "step": 43
     },
     {
       "epoch": 0.031216743526073076,
-      "grad_norm": 1.4895109510972315,
+      "grad_norm": 1.9610946251979264,
       "learning_rate": 9.148936170212767e-06,
-      "loss": 0.1603,
+      "loss": 0.1614,
       "step": 44
     },
     {
       "epoch": 0.03192621496984746,
-      "grad_norm": 1.7589328937210027,
+      "grad_norm": 3.806116355988534,
       "learning_rate": 9.361702127659576e-06,
       "loss": 0.1649,
       "step": 45
     },
     {
       "epoch": 0.03263568641362185,
-      "grad_norm": 7.489278163758361,
+      "grad_norm": 1.6308337200888532,
       "learning_rate": 9.574468085106385e-06,
-      "loss": 0.1658,
+      "loss": 0.1528,
       "step": 46
     },
     {
       "epoch": 0.03334515785739624,
-      "grad_norm": 6.813697836410797,
+      "grad_norm": 1.0471741398584304,
       "learning_rate": 9.787234042553192e-06,
-      "loss": 0.1661,
+      "loss": 0.151,
       "step": 47
     },
     {
       "epoch": 0.034054629301170626,
-      "grad_norm": 4.879033493658327,
+      "grad_norm": 2.1098801420979885,
       "learning_rate": 9.999999999999999e-06,
-      "loss": 0.1618,
+      "loss": 0.15,
       "step": 48
     },
     {
       "epoch": 0.034764100744945015,
-      "grad_norm": 1.7876593798959437,
+      "grad_norm": 1.4498470002731845,
       "learning_rate": 1.0212765957446808e-05,
-      "loss": 0.1404,
+      "loss": 0.1385,
       "step": 49
     },
     {
       "epoch": 0.0354735721887194,
-      "grad_norm": 5.047192381775066,
+      "grad_norm": 2.4843540113556632,
       "learning_rate": 1.0425531914893617e-05,
-      "loss": 0.1692,
+      "loss": 0.1822,
       "step": 50
+    },
+    {
+      "epoch": 0.03618304363249379,
+      "grad_norm": 4.7022928228876335,
+      "learning_rate": 1.0638297872340426e-05,
+      "loss": 0.1465,
+      "step": 51
+    },
+    {
+      "epoch": 0.03689251507626818,
+      "grad_norm": 3.6925599893471666,
+      "learning_rate": 1.0851063829787235e-05,
+      "loss": 0.1301,
+      "step": 52
+    },
+    {
+      "epoch": 0.03760198652004257,
+      "grad_norm": 1.680217688963427,
+      "learning_rate": 1.1063829787234042e-05,
+      "loss": 0.1495,
+      "step": 53
+    },
+    {
+      "epoch": 0.03831145796381696,
+      "grad_norm": 7.334120820499479,
+      "learning_rate": 1.1276595744680851e-05,
+      "loss": 0.1686,
+      "step": 54
+    },
+    {
+      "epoch": 0.039020929407591345,
+      "grad_norm": 3.0183779232825234,
+      "learning_rate": 1.148936170212766e-05,
+      "loss": 0.1567,
+      "step": 55
+    },
+    {
+      "epoch": 0.039730400851365734,
+      "grad_norm": 1.2525087005082511,
+      "learning_rate": 1.170212765957447e-05,
+      "loss": 0.1444,
+      "step": 56
+    },
+    {
+      "epoch": 0.04043987229514012,
+      "grad_norm": 1.425039005914065,
+      "learning_rate": 1.1914893617021277e-05,
+      "loss": 0.1484,
+      "step": 57
+    },
+    {
+      "epoch": 0.04114934373891451,
+      "grad_norm": 2.1985450386586933,
+      "learning_rate": 1.2127659574468084e-05,
+      "loss": 0.1502,
+      "step": 58
+    },
+    {
+      "epoch": 0.0418588151826889,
+      "grad_norm": 3.249197889697388,
+      "learning_rate": 1.2340425531914893e-05,
+      "loss": 0.1473,
+      "step": 59
+    },
+    {
+      "epoch": 0.04256828662646329,
+      "grad_norm": 2.0906804833194443,
+      "learning_rate": 1.2553191489361702e-05,
+      "loss": 0.1415,
+      "step": 60
+    },
+    {
+      "epoch": 0.043277758070237676,
+      "grad_norm": 6.008820861881305,
+      "learning_rate": 1.2765957446808511e-05,
+      "loss": 0.1528,
+      "step": 61
+    },
+    {
+      "epoch": 0.04398722951401206,
+      "grad_norm": 3.510113211053961,
+      "learning_rate": 1.297872340425532e-05,
+      "loss": 0.1659,
+      "step": 62
+    },
+    {
+      "epoch": 0.044696700957786446,
+      "grad_norm": 3.310262198116976,
+      "learning_rate": 1.3191489361702129e-05,
+      "loss": 0.1318,
+      "step": 63
+    },
+    {
+      "epoch": 0.045406172401560835,
+      "grad_norm": 2.998742157037864,
+      "learning_rate": 1.3404255319148936e-05,
+      "loss": 0.1444,
+      "step": 64
+    },
+    {
+      "epoch": 0.04611564384533522,
+      "grad_norm": 3.037454924126702,
+      "learning_rate": 1.3617021276595745e-05,
+      "loss": 0.1534,
+      "step": 65
+    },
+    {
+      "epoch": 0.04682511528910961,
+      "grad_norm": 3.0210908271546746,
+      "learning_rate": 1.3829787234042554e-05,
+      "loss": 0.1402,
+      "step": 66
+    },
+    {
+      "epoch": 0.047534586732884,
+      "grad_norm": 5.520540860172146,
+      "learning_rate": 1.4042553191489362e-05,
+      "loss": 0.1425,
+      "step": 67
+    },
+    {
+      "epoch": 0.04824405817665839,
+      "grad_norm": 2.8578795762705336,
+      "learning_rate": 1.425531914893617e-05,
+      "loss": 0.1441,
+      "step": 68
+    },
+    {
+      "epoch": 0.04895352962043278,
+      "grad_norm": 3.0093954042315523,
+      "learning_rate": 1.4468085106382978e-05,
+      "loss": 0.1209,
+      "step": 69
+    },
+    {
+      "epoch": 0.049663001064207166,
+      "grad_norm": 4.7761651201536,
+      "learning_rate": 1.4680851063829787e-05,
+      "loss": 0.1368,
+      "step": 70
+    },
+    {
+      "epoch": 0.050372472507981554,
+      "grad_norm": 1.479679010072238,
+      "learning_rate": 1.4893617021276596e-05,
+      "loss": 0.1228,
+      "step": 71
+    },
+    {
+      "epoch": 0.05108194395175594,
+      "grad_norm": 1.4562601345388664,
+      "learning_rate": 1.5106382978723403e-05,
+      "loss": 0.1504,
+      "step": 72
+    },
+    {
+      "epoch": 0.05179141539553033,
+      "grad_norm": 1.008711424425401,
+      "learning_rate": 1.531914893617021e-05,
+      "loss": 0.1354,
+      "step": 73
+    },
+    {
+      "epoch": 0.05250088683930472,
+      "grad_norm": 1.1921518196444796,
+      "learning_rate": 1.553191489361702e-05,
+      "loss": 0.1272,
+      "step": 74
+    },
+    {
+      "epoch": 0.05321035828307911,
+      "grad_norm": 2.3190578291261152,
+      "learning_rate": 1.574468085106383e-05,
+      "loss": 0.1307,
+      "step": 75
+    },
+    {
+      "epoch": 0.053919829726853497,
+      "grad_norm": 0.8942079530943704,
+      "learning_rate": 1.5957446808510637e-05,
+      "loss": 0.1215,
+      "step": 76
+    },
+    {
+      "epoch": 0.054629301170627885,
+      "grad_norm": 2.0576355533811554,
+      "learning_rate": 1.6170212765957446e-05,
+      "loss": 0.1362,
+      "step": 77
+    },
+    {
+      "epoch": 0.05533877261440227,
+      "grad_norm": 2.2622110548173677,
+      "learning_rate": 1.6382978723404255e-05,
+      "loss": 0.1899,
+      "step": 78
+    },
+    {
+      "epoch": 0.056048244058176655,
+      "grad_norm": 2.374559556727405,
+      "learning_rate": 1.6595744680851064e-05,
+      "loss": 0.1611,
+      "step": 79
+    },
+    {
+      "epoch": 0.056757715501951043,
+      "grad_norm": 5.447170725571408,
+      "learning_rate": 1.6808510638297873e-05,
+      "loss": 0.1514,
+      "step": 80
+    },
+    {
+      "epoch": 0.05746718694572543,
+      "grad_norm": 2.5429748228117814,
+      "learning_rate": 1.7021276595744682e-05,
+      "loss": 0.1468,
+      "step": 81
+    },
+    {
+      "epoch": 0.05817665838949982,
+      "grad_norm": 2.7415968193046902,
+      "learning_rate": 1.723404255319149e-05,
+      "loss": 0.1445,
+      "step": 82
+    },
+    {
+      "epoch": 0.05888612983327421,
+      "grad_norm": 3.0382016331376094,
+      "learning_rate": 1.7446808510638297e-05,
+      "loss": 0.158,
+      "step": 83
+    },
+    {
+      "epoch": 0.0595956012770486,
+      "grad_norm": 1.6325149171380335,
+      "learning_rate": 1.7659574468085106e-05,
+      "loss": 0.1253,
+      "step": 84
+    },
+    {
+      "epoch": 0.060305072720822986,
+      "grad_norm": 3.316327985423067,
+      "learning_rate": 1.7872340425531915e-05,
+      "loss": 0.1418,
+      "step": 85
+    },
+    {
+      "epoch": 0.061014544164597374,
+      "grad_norm": 6.588395895312288,
+      "learning_rate": 1.8085106382978724e-05,
+      "loss": 0.1455,
+      "step": 86
+    },
+    {
+      "epoch": 0.06172401560837176,
+      "grad_norm": 1.5641033467639194,
+      "learning_rate": 1.8297872340425533e-05,
+      "loss": 0.139,
+      "step": 87
+    },
+    {
+      "epoch": 0.06243348705214615,
+      "grad_norm": 1.6297579083033085,
+      "learning_rate": 1.8510638297872342e-05,
+      "loss": 0.1396,
+      "step": 88
+    },
+    {
+      "epoch": 0.06314295849592054,
+      "grad_norm": 6.970523654466422,
+      "learning_rate": 1.872340425531915e-05,
+      "loss": 0.1568,
+      "step": 89
+    },
+    {
+      "epoch": 0.06385242993969492,
+      "grad_norm": 1.271092780714167,
+      "learning_rate": 1.893617021276596e-05,
+      "loss": 0.147,
+      "step": 90
+    },
+    {
+      "epoch": 0.06456190138346932,
+      "grad_norm": 9.422206811592059,
+      "learning_rate": 1.914893617021277e-05,
+      "loss": 0.1603,
+      "step": 91
+    },
+    {
+      "epoch": 0.0652713728272437,
+      "grad_norm": 6.043928089777284,
+      "learning_rate": 1.9361702127659575e-05,
+      "loss": 0.1731,
+      "step": 92
+    },
+    {
+      "epoch": 0.0659808442710181,
+      "grad_norm": 1.6440555166858408,
+      "learning_rate": 1.9574468085106384e-05,
+      "loss": 0.1323,
+      "step": 93
+    },
+    {
+      "epoch": 0.06669031571479248,
+      "grad_norm": 3.722316535878048,
+      "learning_rate": 1.978723404255319e-05,
+      "loss": 0.123,
+      "step": 94
+    },
+    {
+      "epoch": 0.06739978715856687,
+      "grad_norm": 0.876265361813727,
+      "learning_rate": 1.9999999999999998e-05,
+      "loss": 0.1492,
+      "step": 95
+    },
+    {
+      "epoch": 0.06810925860234125,
+      "grad_norm": 2.9790618574403775,
+      "learning_rate": 2.0212765957446807e-05,
+      "loss": 0.1422,
+      "step": 96
+    },
+    {
+      "epoch": 0.06881873004611565,
+      "grad_norm": 4.427704225480461,
+      "learning_rate": 2.0425531914893616e-05,
+      "loss": 0.1263,
+      "step": 97
+    },
+    {
+      "epoch": 0.06952820148989003,
+      "grad_norm": 2.574380283469407,
+      "learning_rate": 2.0638297872340425e-05,
+      "loss": 0.1027,
+      "step": 98
+    },
+    {
+      "epoch": 0.07023767293366442,
+      "grad_norm": 2.1421248571266838,
+      "learning_rate": 2.0851063829787234e-05,
+      "loss": 0.1315,
+      "step": 99
+    },
+    {
+      "epoch": 0.0709471443774388,
+      "grad_norm": 5.305110629941279,
+      "learning_rate": 2.1063829787234043e-05,
+      "loss": 0.1971,
+      "step": 100
     }
   ],
   "logging_steps": 1,
@@ -377,7 +727,7 @@
       "attributes": {}
     }
   },
-  "total_flos": 31813017403392.0,
+  "total_flos": 64740426121216.0,
   "train_batch_size": 10,
   "trial_name": null,
   "trial_params": null

last-checkpoint/training_args.bin

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:59690dec2de0315e951e84e970a3e21faeab8de9ac11de6607c8728b4bc47b50
-size 8017
+oid sha256:def4d75ea319d525641ec3446da18695a7865c879bc733e7cf5fa86aa02e3106
+size 7953