Training in progress, step 20000, checkpoint

checkpoint-20000/model.safetensors

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:7c5010e5bb985d46725bc14def182e65eb2df32dc603252a77ed51ccb9e16c38
+oid sha256:aa8bdcb72d4cf420be6e6db10f6ba46f48d9d180adf71a6184b55adbb78696c9
 size 14809584

checkpoint-20000/optimizer.pt

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:aabcb11de07b958d88e22d5e88cdf756709f34147fa3787bb02be53f32cb23f7
+oid sha256:92b81b7829420cff77b4513764357686cc35a98bcebd86cc43604403217148c0
 size 29680378

checkpoint-20000/scheduler.pt

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:19c8d50d5c4f433e4641aae9ea0af9860b8650749edbcbbf916547da1333f81e
+oid sha256:e1ed21e6f749fb2fd1b7e2e44a48da2ce954d40934096e439c96e70e0947df19
 size 1192

checkpoint-20000/tokenizer.json

@@ -1,21 +1,7 @@
 {
   "version": "1.0",
-  "truncation": {
-    "direction": "Right",
-    "max_length": 128,
-    "strategy": "LongestFirst",
-    "stride": 0
-  },
-  "padding": {
-    "strategy": {
-      "Fixed": 128
-    },
-    "direction": "Right",
-    "pad_to_multiple_of": null,
-    "pad_id": 3,
-    "pad_type_id": 0,
-    "pad_token": "[PAD]"
-  },
+  "truncation": null,
+  "padding": null,
   "added_tokens": [
     {
       "id": 0,

checkpoint-20000/trainer_state.json

@@ -1,7 +1,7 @@
 {
-  "best_metric": 0.15438680350780487,
+  "best_metric": 3.5121970176696777,
   "best_model_checkpoint": "/Users/frapadovani/Desktop/babyLM_controlled/models_trained/convergence_french/random_sentence_french/checkpoint-20000",
-  "epoch": 0.03616819658138206,
+  "epoch": 0.5921890267373345,
   "eval_steps": 2000,
   "global_step": 20000,
   "is_hyper_param_search": false,
@@ -9,158 +9,158 @@
   "is_world_process_zero": true,
   "log_history": [
     {
-      "epoch": 0.003616819658138206,
-      "grad_norm": 0.14744240045547485,
+      "epoch": 0.059218902673733455,
+      "grad_norm": 1.0812722444534302,
       "learning_rate": 0.0001,
-      "loss": 0.7697,
+      "loss": 4.7523,
       "step": 2000
     },
     {
-      "epoch": 0.003616819658138206,
-      "eval_loss": 0.19154727458953857,
-      "eval_runtime": 91.812,
-      "eval_samples_per_second": 397.933,
-      "eval_steps_per_second": 24.877,
+      "epoch": 0.059218902673733455,
+      "eval_loss": 4.0236735343933105,
+      "eval_runtime": 5.173,
+      "eval_samples_per_second": 425.477,
+      "eval_steps_per_second": 26.677,
       "step": 2000
     },
     {
-      "epoch": 0.007233639316276412,
-      "grad_norm": 0.1623348593711853,
+      "epoch": 0.11843780534746691,
+      "grad_norm": 1.2557106018066406,
       "learning_rate": 0.0001,
-      "loss": 0.1884,
+      "loss": 3.8958,
       "step": 4000
     },
     {
-      "epoch": 0.007233639316276412,
-      "eval_loss": 0.17771275341510773,
-      "eval_runtime": 99.0721,
-      "eval_samples_per_second": 368.772,
-      "eval_steps_per_second": 23.054,
+      "epoch": 0.11843780534746691,
+      "eval_loss": 3.8087611198425293,
+      "eval_runtime": 5.2666,
+      "eval_samples_per_second": 417.913,
+      "eval_steps_per_second": 26.203,
       "step": 4000
     },
     {
-      "epoch": 0.010850458974414617,
-      "grad_norm": 0.13657265901565552,
+      "epoch": 0.17765670802120037,
+      "grad_norm": 1.3326870203018188,
       "learning_rate": 0.0001,
-      "loss": 0.178,
+      "loss": 3.7344,
       "step": 6000
     },
     {
-      "epoch": 0.010850458974414617,
-      "eval_loss": 0.16984443366527557,
-      "eval_runtime": 90.3285,
-      "eval_samples_per_second": 404.468,
-      "eval_steps_per_second": 25.285,
+      "epoch": 0.17765670802120037,
+      "eval_loss": 3.704986333847046,
+      "eval_runtime": 5.2277,
+      "eval_samples_per_second": 421.023,
+      "eval_steps_per_second": 26.398,
       "step": 6000
     },
     {
-      "epoch": 0.014467278632552824,
-      "grad_norm": 0.15028506517410278,
+      "epoch": 0.23687561069493382,
+      "grad_norm": 1.38993239402771,
       "learning_rate": 0.0001,
-      "loss": 0.1714,
+      "loss": 3.6379,
       "step": 8000
     },
     {
-      "epoch": 0.014467278632552824,
-      "eval_loss": 0.16548627614974976,
-      "eval_runtime": 90.9787,
-      "eval_samples_per_second": 401.578,
-      "eval_steps_per_second": 25.105,
+      "epoch": 0.23687561069493382,
+      "eval_loss": 3.6455252170562744,
+      "eval_runtime": 5.1485,
+      "eval_samples_per_second": 427.5,
+      "eval_steps_per_second": 26.804,
       "step": 8000
     },
     {
-      "epoch": 0.01808409829069103,
-      "grad_norm": 0.17982865869998932,
+      "epoch": 0.29609451336866727,
+      "grad_norm": 1.3884963989257812,
       "learning_rate": 0.0001,
-      "loss": 0.1669,
+      "loss": 3.5687,
       "step": 10000
     },
     {
-      "epoch": 0.01808409829069103,
-      "eval_loss": 0.16203773021697998,
-      "eval_runtime": 87.7097,
-      "eval_samples_per_second": 416.544,
-      "eval_steps_per_second": 26.04,
+      "epoch": 0.29609451336866727,
+      "eval_loss": 3.6062421798706055,
+      "eval_runtime": 5.1466,
+      "eval_samples_per_second": 427.661,
+      "eval_steps_per_second": 26.814,
       "step": 10000
     },
     {
-      "epoch": 0.021700917948829234,
-      "grad_norm": 0.14700624346733093,
+      "epoch": 0.35531341604240074,
+      "grad_norm": 1.435062289237976,
       "learning_rate": 0.0001,
-      "loss": 0.1647,
+      "loss": 3.5165,
       "step": 12000
     },
     {
-      "epoch": 0.021700917948829234,
-      "eval_loss": 0.1595447063446045,
-      "eval_runtime": 87.4086,
-      "eval_samples_per_second": 417.979,
-      "eval_steps_per_second": 26.13,
+      "epoch": 0.35531341604240074,
+      "eval_loss": 3.5764389038085938,
+      "eval_runtime": 5.132,
+      "eval_samples_per_second": 428.881,
+      "eval_steps_per_second": 26.89,
       "step": 12000
     },
     {
-      "epoch": 0.02531773760696744,
-      "grad_norm": 0.1563248634338379,
+      "epoch": 0.41453231871613416,
+      "grad_norm": 1.3539327383041382,
       "learning_rate": 0.0001,
-      "loss": 0.1608,
+      "loss": 3.4738,
       "step": 14000
     },
     {
-      "epoch": 0.02531773760696744,
-      "eval_loss": 0.1577664017677307,
-      "eval_runtime": 87.7156,
-      "eval_samples_per_second": 416.517,
-      "eval_steps_per_second": 26.039,
+      "epoch": 0.41453231871613416,
+      "eval_loss": 3.555619239807129,
+      "eval_runtime": 5.1586,
+      "eval_samples_per_second": 426.667,
+      "eval_steps_per_second": 26.752,
       "step": 14000
     },
     {
-      "epoch": 0.02893455726510565,
-      "grad_norm": 0.1511106789112091,
+      "epoch": 0.47375122138986764,
+      "grad_norm": 1.3827875852584839,
       "learning_rate": 0.0001,
-      "loss": 0.1601,
+      "loss": 3.4389,
       "step": 16000
     },
     {
-      "epoch": 0.02893455726510565,
-      "eval_loss": 0.15645398199558258,
-      "eval_runtime": 90.1398,
-      "eval_samples_per_second": 405.315,
-      "eval_steps_per_second": 25.338,
+      "epoch": 0.47375122138986764,
+      "eval_loss": 3.5388290882110596,
+      "eval_runtime": 5.8029,
+      "eval_samples_per_second": 379.292,
+      "eval_steps_per_second": 23.781,
       "step": 16000
     },
     {
-      "epoch": 0.032551376923243856,
-      "grad_norm": 0.16017386317253113,
+      "epoch": 0.5329701240636011,
+      "grad_norm": 1.316589593887329,
       "learning_rate": 0.0001,
-      "loss": 0.1607,
+      "loss": 3.4067,
       "step": 18000
     },
     {
-      "epoch": 0.032551376923243856,
-      "eval_loss": 0.1552029848098755,
-      "eval_runtime": 87.911,
-      "eval_samples_per_second": 415.591,
-      "eval_steps_per_second": 25.981,
+      "epoch": 0.5329701240636011,
+      "eval_loss": 3.5234413146972656,
+      "eval_runtime": 5.1383,
+      "eval_samples_per_second": 428.353,
+      "eval_steps_per_second": 26.857,
       "step": 18000
     },
     {
-      "epoch": 0.03616819658138206,
-      "grad_norm": 0.12754733860492706,
+      "epoch": 0.5921890267373345,
+      "grad_norm": 1.3910475969314575,
       "learning_rate": 0.0001,
-      "loss": 0.1565,
+      "loss": 3.3778,
       "step": 20000
     },
     {
-      "epoch": 0.03616819658138206,
-      "eval_loss": 0.15438680350780487,
-      "eval_runtime": 87.7234,
-      "eval_samples_per_second": 416.479,
-      "eval_steps_per_second": 26.036,
+      "epoch": 0.5921890267373345,
+      "eval_loss": 3.5121970176696777,
+      "eval_runtime": 7.6572,
+      "eval_samples_per_second": 287.442,
+      "eval_steps_per_second": 18.022,
       "step": 20000
     }
   ],
   "logging_steps": 2000,
-  "max_steps": 552972,
+  "max_steps": 33773,
   "num_input_tokens_seen": 0,
   "num_train_epochs": 1,
   "save_steps": 2000,
@@ -171,7 +171,7 @@
         "early_stopping_threshold": 0.001
       },
       "attributes": {
-        "early_stopping_patience_counter": 1
+        "early_stopping_patience_counter": 0
       }
     },
     "TrainerControl": {

checkpoint-20000/training_args.bin

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:f5e294196a968439ae6368c7b1ef6eb6e2124702ff760349c5ba2ea15623271d
+oid sha256:fce3a8508ad2aa1c0ae1cc54c6fe1747e2312953499cad9c035bc201c305894e
 size 5496