Training in progress, step 24000, checkpoint

checkpoint-24000/model.safetensors

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:2323f734790f984460f10ce290a07ed2fdee95037bd31b9f168e82286e89993e
+oid sha256:57fb9bc470b9dbc3a3608c1af2dd6b51f35dee546dc68b853ef9177f24a598b5
 size 51007160

checkpoint-24000/optimizer.pt

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:6837ac61e2c391db7039a5c9ccdf7accff1772b83874d3670054d2935fe147e0
+oid sha256:62120f961497f03f065bf4f19d428e36d741e324bab29d13162e54022820a662
 size 102078202

checkpoint-24000/rng_state.pth

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:0310d60d4d914a6ddd487bf19ff128fe8584de8a19e727db4b59e97e1421e293
+oid sha256:d3c6b91a99c8904150e8b7b2af46162c45ddbecb227786c820af530ae62d24c4
 size 14308

checkpoint-24000/scheduler.pt

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:0d39773d3586692cd5bc5596907a4d2bdae72ffeb1c423c71a80d3870f87e15a
+oid sha256:77024f61a7926e2703da5222cefed4fa0e7a12ad732072f14ea0d56db7394fc3
 size 1000

checkpoint-24000/trainer_state.json

@@ -1,7 +1,7 @@
 {
-  "best_metric": 4.683297634124756,
-  "best_model_checkpoint": "/home/p318482/babyLM_controlled/models_trained/en_clm/wikipedia_30/checkpoint-24000",
-  "epoch": 12.621614514856692,
+  "best_metric": 3.7359092235565186,
+  "best_model_checkpoint": "/home/p318482/babyLM_controlled/models_trained/fr_clm/wikipedia_30/checkpoint-24000",
+  "epoch": 25.764895330112722,
   "eval_steps": 2000,
   "global_step": 24000,
   "is_hyper_param_search": false,
@@ -9,148 +9,148 @@
   "is_world_process_zero": true,
   "log_history": [
     {
-      "epoch": 1.0518012095713911,
-      "eval_loss": 7.518920421600342,
-      "eval_runtime": 2.2399,
-      "eval_samples_per_second": 1422.808,
-      "eval_steps_per_second": 89.288,
+      "epoch": 2.1470746108427265,
+      "eval_loss": 7.138043403625488,
+      "eval_runtime": 0.8415,
+      "eval_samples_per_second": 1273.93,
+      "eval_steps_per_second": 79.621,
       "step": 2000
     },
     {
-      "epoch": 2.1036024191427822,
-      "grad_norm": 1.2947496175765991,
+      "epoch": 4.294149221685453,
+      "grad_norm": 1.455039381980896,
       "learning_rate": 1e-05,
-      "loss": 7.6141,
+      "loss": 7.213,
       "step": 4000
     },
     {
-      "epoch": 2.1036024191427822,
-      "eval_loss": 6.568221092224121,
-      "eval_runtime": 2.1236,
-      "eval_samples_per_second": 1500.783,
-      "eval_steps_per_second": 94.182,
+      "epoch": 4.294149221685453,
+      "eval_loss": 5.8541717529296875,
+      "eval_runtime": 0.7753,
+      "eval_samples_per_second": 1382.676,
+      "eval_steps_per_second": 86.417,
       "step": 4000
     },
     {
-      "epoch": 3.155403628714173,
-      "eval_loss": 6.16925573348999,
-      "eval_runtime": 2.1624,
-      "eval_samples_per_second": 1473.825,
-      "eval_steps_per_second": 92.49,
+      "epoch": 6.4412238325281805,
+      "eval_loss": 5.403579235076904,
+      "eval_runtime": 0.7542,
+      "eval_samples_per_second": 1421.437,
+      "eval_steps_per_second": 88.84,
       "step": 6000
     },
     {
-      "epoch": 4.2072048382855645,
-      "grad_norm": 2.330080270767212,
-      "learning_rate": 2e-05,
-      "loss": 6.2396,
+      "epoch": 8.588298443370906,
+      "grad_norm": 3.117489814758301,
+      "learning_rate": 1.9997500000000003e-05,
+      "loss": 5.4304,
       "step": 8000
     },
     {
-      "epoch": 4.2072048382855645,
-      "eval_loss": 5.901575088500977,
-      "eval_runtime": 2.1326,
-      "eval_samples_per_second": 1494.388,
-      "eval_steps_per_second": 93.78,
+      "epoch": 8.588298443370906,
+      "eval_loss": 5.049880504608154,
+      "eval_runtime": 0.7597,
+      "eval_samples_per_second": 1411.127,
+      "eval_steps_per_second": 88.195,
       "step": 8000
     },
     {
-      "epoch": 5.259006047856955,
-      "eval_loss": 5.67369270324707,
-      "eval_runtime": 2.1364,
-      "eval_samples_per_second": 1491.788,
-      "eval_steps_per_second": 93.617,
+      "epoch": 10.735373054213634,
+      "eval_loss": 4.760603427886963,
+      "eval_runtime": 0.7624,
+      "eval_samples_per_second": 1406.053,
+      "eval_steps_per_second": 87.878,
       "step": 10000
     },
     {
-      "epoch": 6.310807257428346,
-      "grad_norm": 2.565220594406128,
-      "learning_rate": 2.99925e-05,
-      "loss": 5.7217,
+      "epoch": 12.882447665056361,
+      "grad_norm": 2.9511141777038574,
+      "learning_rate": 2.9995e-05,
+      "loss": 4.771,
       "step": 12000
     },
     {
-      "epoch": 6.310807257428346,
-      "eval_loss": 5.460005760192871,
-      "eval_runtime": 2.1835,
-      "eval_samples_per_second": 1459.563,
-      "eval_steps_per_second": 91.595,
+      "epoch": 12.882447665056361,
+      "eval_loss": 4.517208576202393,
+      "eval_runtime": 0.764,
+      "eval_samples_per_second": 1403.097,
+      "eval_steps_per_second": 87.694,
       "step": 12000
     },
     {
-      "epoch": 7.362608466999737,
-      "eval_loss": 5.272489547729492,
-      "eval_runtime": 2.1833,
-      "eval_samples_per_second": 1459.691,
-      "eval_steps_per_second": 91.603,
+      "epoch": 15.029522275899087,
+      "eval_loss": 4.3206024169921875,
+      "eval_runtime": 0.7595,
+      "eval_samples_per_second": 1411.435,
+      "eval_steps_per_second": 88.215,
       "step": 14000
     },
     {
-      "epoch": 8.414409676571129,
-      "grad_norm": 2.7334115505218506,
-      "learning_rate": 3.99875e-05,
-      "loss": 5.3064,
+      "epoch": 17.176596886741816,
+      "grad_norm": 2.955515146255493,
+      "learning_rate": 3.999e-05,
+      "loss": 4.2888,
       "step": 16000
     },
     {
-      "epoch": 8.414409676571129,
-      "eval_loss": 5.1074652671813965,
-      "eval_runtime": 2.2402,
-      "eval_samples_per_second": 1422.652,
-      "eval_steps_per_second": 89.278,
+      "epoch": 17.176596886741816,
+      "eval_loss": 4.152973175048828,
+      "eval_runtime": 0.8247,
+      "eval_samples_per_second": 1299.91,
+      "eval_steps_per_second": 81.244,
       "step": 16000
     },
     {
-      "epoch": 9.46621088614252,
-      "eval_loss": 4.9716596603393555,
-      "eval_runtime": 2.1736,
-      "eval_samples_per_second": 1466.258,
-      "eval_steps_per_second": 92.015,
+      "epoch": 19.32367149758454,
+      "eval_loss": 4.015476703643799,
+      "eval_runtime": 0.7501,
+      "eval_samples_per_second": 1429.237,
+      "eval_steps_per_second": 89.327,
       "step": 18000
     },
     {
-      "epoch": 10.51801209571391,
-      "grad_norm": 2.6722376346588135,
-      "learning_rate": 4.9985e-05,
-      "loss": 4.9744,
+      "epoch": 21.470746108427267,
+      "grad_norm": 3.0531320571899414,
+      "learning_rate": 4.99875e-05,
+      "loss": 3.9141,
       "step": 20000
     },
     {
-      "epoch": 10.51801209571391,
-      "eval_loss": 4.864916801452637,
-      "eval_runtime": 2.2031,
-      "eval_samples_per_second": 1446.585,
-      "eval_steps_per_second": 90.78,
+      "epoch": 21.470746108427267,
+      "eval_loss": 3.896575927734375,
+      "eval_runtime": 0.768,
+      "eval_samples_per_second": 1395.843,
+      "eval_steps_per_second": 87.24,
       "step": 20000
     },
     {
-      "epoch": 11.569813305285301,
-      "eval_loss": 4.763542652130127,
-      "eval_runtime": 2.4382,
-      "eval_samples_per_second": 1307.131,
-      "eval_steps_per_second": 82.029,
+      "epoch": 23.617820719269993,
+      "eval_loss": 3.8046905994415283,
+      "eval_runtime": 0.7601,
+      "eval_samples_per_second": 1410.272,
+      "eval_steps_per_second": 88.142,
       "step": 22000
     },
     {
-      "epoch": 12.621614514856692,
-      "grad_norm": 2.4539408683776855,
-      "learning_rate": 5.9985e-05,
-      "loss": 4.7273,
+      "epoch": 25.764895330112722,
+      "grad_norm": 3.089989185333252,
+      "learning_rate": 5.998250000000001e-05,
+      "loss": 3.6154,
       "step": 24000
     },
     {
-      "epoch": 12.621614514856692,
-      "eval_loss": 4.683297634124756,
-      "eval_runtime": 2.452,
-      "eval_samples_per_second": 1299.733,
-      "eval_steps_per_second": 81.565,
+      "epoch": 25.764895330112722,
+      "eval_loss": 3.7359092235565186,
+      "eval_runtime": 0.7778,
+      "eval_samples_per_second": 1378.238,
+      "eval_steps_per_second": 86.14,
       "step": 24000
     }
   ],
   "logging_steps": 4000,
   "max_steps": 100000,
   "num_input_tokens_seen": 0,
-  "num_train_epochs": 53,
+  "num_train_epochs": 108,
   "save_steps": 4000,
   "stateful_callbacks": {
     "TrainerControl": {
@@ -164,7 +164,7 @@
       "attributes": {}
     }
   },
-  "total_flos": 6204129776286720.0,
+  "total_flos": 6203974169395200.0,
   "train_batch_size": 16,
   "trial_name": null,
   "trial_params": null

checkpoint-24000/training_args.bin

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:ab68434ec7156d3a63e4783f824871a48d62c7c2fdcc831fdd5471c5f4aec7fe
+oid sha256:26102d0ac750ce531ebda9d7577ccea6c8ecbc93b04bdff23226e376dd5609a2
 size 5368