End of training

README.md

@@ -3,6 +3,9 @@ library_name: transformers
 license: cc-by-nc-4.0
 base_model: facebook/mms-1b-all
 tags:
+- automatic-speech-recognition
+- bigcgen
+- mms
 - generated_from_trainer
 metrics:
 - wer
@@ -16,9 +19,9 @@ should probably proofread and complete it, then remove this comment. -->
 
 # mms-1b-bigcgen-male-5hrs-model
 
-This model is a fine-tuned version of [facebook/mms-1b-all](https://huggingface.co/facebook/mms-1b-all) on an unknown dataset.
+This model is a fine-tuned version of [facebook/mms-1b-all](https://huggingface.co/facebook/mms-1b-all) on the BIGCGEN - BEM dataset.
 It achieves the following results on the evaluation set:
-- Loss: 0.4407
+- Loss: 0.4408
 - Wer: 0.4520
 
 ## Model description

adapter.bem.safetensors

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:9140743aad5dec810fbd36760fcca33e9cadf20736ddb460086e654ddac60d85
+size 8788284

all_results.json

@@ -0,0 +1,15 @@
+{
+    "epoch": 7.763975155279503,
+    "eval_loss": 0.4407535195350647,
+    "eval_runtime": 32.0903,
+    "eval_samples": 441,
+    "eval_samples_per_second": 13.742,
+    "eval_steps_per_second": 3.459,
+    "eval_wer": 0.4520317383986535,
+    "total_flos": 1.301164200639248e+19,
+    "train_loss": 1.6717456512451172,
+    "train_runtime": 3684.5977,
+    "train_samples": 2573,
+    "train_samples_per_second": 20.949,
+    "train_steps_per_second": 2.622
+}

eval_results.json

@@ -0,0 +1,9 @@
+{
+    "epoch": 7.763975155279503,
+    "eval_loss": 0.4407535195350647,
+    "eval_runtime": 32.0903,
+    "eval_samples": 441,
+    "eval_samples_per_second": 13.742,
+    "eval_steps_per_second": 3.459,
+    "eval_wer": 0.4520317383986535
+}

runs/Dec27_22-45-03_srvrocgpu011.uct.ac.za/events.out.tfevents.1735336448.srvrocgpu011.uct.ac.za

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:d6fd52e59cdb83c772e367b0e15e5061fe1d2307a768020897ab179d39861f56
+size 40

train_results.json

@@ -0,0 +1,9 @@
+{
+    "epoch": 7.763975155279503,
+    "total_flos": 1.301164200639248e+19,
+    "train_loss": 1.6717456512451172,
+    "train_runtime": 3684.5977,
+    "train_samples": 2573,
+    "train_samples_per_second": 20.949,
+    "train_steps_per_second": 2.622
+}

trainer_state.json

@@ -0,0 +1,451 @@
+{
+  "best_metric": 0.4343813359737396,
+  "best_model_checkpoint": "/scratch/skscla001/speech/results/mms-1b-bigcgen-male-5hrs-model/checkpoint-2100",
+  "epoch": 7.763975155279503,
+  "eval_steps": 100,
+  "global_step": 2500,
+  "is_hyper_param_search": false,
+  "is_local_process_zero": true,
+  "is_world_process_zero": true,
+  "log_history": [
+    {
+      "epoch": 0.3105590062111801,
+      "grad_norm": 7.037683486938477,
+      "learning_rate": 0.00028799999999999995,
+      "loss": 12.4451,
+      "step": 100
+    },
+    {
+      "epoch": 0.3105590062111801,
+      "eval_loss": 1.20354163646698,
+      "eval_runtime": 31.8229,
+      "eval_samples_per_second": 13.858,
+      "eval_steps_per_second": 3.488,
+      "eval_wer": 0.8328925222409232,
+      "step": 100
+    },
+    {
+      "epoch": 0.6211180124223602,
+      "grad_norm": 3.7343695163726807,
+      "learning_rate": 0.00029698744769874474,
+      "loss": 1.6342,
+      "step": 200
+    },
+    {
+      "epoch": 0.6211180124223602,
+      "eval_loss": 0.6174936890602112,
+      "eval_runtime": 31.7812,
+      "eval_samples_per_second": 13.876,
+      "eval_steps_per_second": 3.493,
+      "eval_wer": 0.5758595816301996,
+      "step": 200
+    },
+    {
+      "epoch": 0.9316770186335404,
+      "grad_norm": 4.2912797927856445,
+      "learning_rate": 0.0002938493723849372,
+      "loss": 1.5403,
+      "step": 300
+    },
+    {
+      "epoch": 0.9316770186335404,
+      "eval_loss": 0.5695027709007263,
+      "eval_runtime": 31.6004,
+      "eval_samples_per_second": 13.956,
+      "eval_steps_per_second": 3.513,
+      "eval_wer": 0.5534984371243087,
+      "step": 300
+    },
+    {
+      "epoch": 1.2422360248447206,
+      "grad_norm": 1.983912467956543,
+      "learning_rate": 0.00029071129707112966,
+      "loss": 1.3778,
+      "step": 400
+    },
+    {
+      "epoch": 1.2422360248447206,
+      "eval_loss": 0.5523655414581299,
+      "eval_runtime": 31.7017,
+      "eval_samples_per_second": 13.911,
+      "eval_steps_per_second": 3.501,
+      "eval_wer": 0.5359461408992546,
+      "step": 400
+    },
+    {
+      "epoch": 1.5527950310559007,
+      "grad_norm": 1.4964157342910767,
+      "learning_rate": 0.00028757322175732215,
+      "loss": 1.4572,
+      "step": 500
+    },
+    {
+      "epoch": 1.5527950310559007,
+      "eval_loss": 0.5301775336265564,
+      "eval_runtime": 32.0399,
+      "eval_samples_per_second": 13.764,
+      "eval_steps_per_second": 3.464,
+      "eval_wer": 0.5171916326039914,
+      "step": 500
+    },
+    {
+      "epoch": 1.8633540372670807,
+      "grad_norm": 3.8727850914001465,
+      "learning_rate": 0.0002844665271966527,
+      "loss": 1.4042,
+      "step": 600
+    },
+    {
+      "epoch": 1.8633540372670807,
+      "eval_loss": 0.5179265737533569,
+      "eval_runtime": 31.8425,
+      "eval_samples_per_second": 13.849,
+      "eval_steps_per_second": 3.486,
+      "eval_wer": 0.5265688867516229,
+      "step": 600
+    },
+    {
+      "epoch": 2.1739130434782608,
+      "grad_norm": 2.505019187927246,
+      "learning_rate": 0.0002813284518828452,
+      "loss": 1.4053,
+      "step": 700
+    },
+    {
+      "epoch": 2.1739130434782608,
+      "eval_loss": 0.5029146671295166,
+      "eval_runtime": 31.795,
+      "eval_samples_per_second": 13.87,
+      "eval_steps_per_second": 3.491,
+      "eval_wer": 0.5143063236354893,
+      "step": 700
+    },
+    {
+      "epoch": 2.4844720496894412,
+      "grad_norm": 2.7380497455596924,
+      "learning_rate": 0.0002781903765690376,
+      "loss": 1.2782,
+      "step": 800
+    },
+    {
+      "epoch": 2.4844720496894412,
+      "eval_loss": 0.4700861871242523,
+      "eval_runtime": 32.0844,
+      "eval_samples_per_second": 13.745,
+      "eval_steps_per_second": 3.46,
+      "eval_wer": 0.4864150036066362,
+      "step": 800
+    },
+    {
+      "epoch": 2.795031055900621,
+      "grad_norm": 1.6105161905288696,
+      "learning_rate": 0.0002750523012552301,
+      "loss": 1.2541,
+      "step": 900
+    },
+    {
+      "epoch": 2.795031055900621,
+      "eval_loss": 0.4585192799568176,
+      "eval_runtime": 32.0616,
+      "eval_samples_per_second": 13.755,
+      "eval_steps_per_second": 3.462,
+      "eval_wer": 0.48665544602067806,
+      "step": 900
+    },
+    {
+      "epoch": 3.1055900621118013,
+      "grad_norm": 3.0428338050842285,
+      "learning_rate": 0.0002719142259414226,
+      "loss": 1.1672,
+      "step": 1000
+    },
+    {
+      "epoch": 3.1055900621118013,
+      "eval_loss": 0.47279417514801025,
+      "eval_runtime": 31.7171,
+      "eval_samples_per_second": 13.904,
+      "eval_steps_per_second": 3.5,
+      "eval_wer": 0.4861745611925944,
+      "step": 1000
+    },
+    {
+      "epoch": 3.4161490683229814,
+      "grad_norm": 1.6724671125411987,
+      "learning_rate": 0.00026877615062761507,
+      "loss": 1.1205,
+      "step": 1100
+    },
+    {
+      "epoch": 3.4161490683229814,
+      "eval_loss": 0.4558369219303131,
+      "eval_runtime": 31.7372,
+      "eval_samples_per_second": 13.895,
+      "eval_steps_per_second": 3.497,
+      "eval_wer": 0.4794421735994229,
+      "step": 1100
+    },
+    {
+      "epoch": 3.7267080745341614,
+      "grad_norm": 4.677685260772705,
+      "learning_rate": 0.0002656380753138075,
+      "loss": 1.1699,
+      "step": 1200
+    },
+    {
+      "epoch": 3.7267080745341614,
+      "eval_loss": 0.4520164430141449,
+      "eval_runtime": 31.9065,
+      "eval_samples_per_second": 13.822,
+      "eval_steps_per_second": 3.479,
+      "eval_wer": 0.4811252704977158,
+      "step": 1200
+    },
+    {
+      "epoch": 4.037267080745342,
+      "grad_norm": 15.624258995056152,
+      "learning_rate": 0.0002625,
+      "loss": 1.2418,
+      "step": 1300
+    },
+    {
+      "epoch": 4.037267080745342,
+      "eval_loss": 0.4494916796684265,
+      "eval_runtime": 32.133,
+      "eval_samples_per_second": 13.724,
+      "eval_steps_per_second": 3.454,
+      "eval_wer": 0.4751142101466699,
+      "step": 1300
+    },
+    {
+      "epoch": 4.3478260869565215,
+      "grad_norm": 1.0827780961990356,
+      "learning_rate": 0.00025936192468619247,
+      "loss": 1.071,
+      "step": 1400
+    },
+    {
+      "epoch": 4.3478260869565215,
+      "eval_loss": 0.4487316310405731,
+      "eval_runtime": 31.845,
+      "eval_samples_per_second": 13.848,
+      "eval_steps_per_second": 3.486,
+      "eval_wer": 0.47367155566241886,
+      "step": 1400
+    },
+    {
+      "epoch": 4.658385093167702,
+      "grad_norm": 2.8134868144989014,
+      "learning_rate": 0.0002562238493723849,
+      "loss": 1.078,
+      "step": 1500
+    },
+    {
+      "epoch": 4.658385093167702,
+      "eval_loss": 0.4445899724960327,
+      "eval_runtime": 31.7402,
+      "eval_samples_per_second": 13.894,
+      "eval_steps_per_second": 3.497,
+      "eval_wer": 0.4760759798028372,
+      "step": 1500
+    },
+    {
+      "epoch": 4.9689440993788825,
+      "grad_norm": 3.0363829135894775,
+      "learning_rate": 0.0002530857740585774,
+      "loss": 1.2474,
+      "step": 1600
+    },
+    {
+      "epoch": 4.9689440993788825,
+      "eval_loss": 0.4437084496021271,
+      "eval_runtime": 32.0173,
+      "eval_samples_per_second": 13.774,
+      "eval_steps_per_second": 3.467,
+      "eval_wer": 0.46261120461649435,
+      "step": 1600
+    },
+    {
+      "epoch": 5.279503105590062,
+      "grad_norm": 1.907182216644287,
+      "learning_rate": 0.0002499476987447699,
+      "loss": 1.1127,
+      "step": 1700
+    },
+    {
+      "epoch": 5.279503105590062,
+      "eval_loss": 0.43802064657211304,
+      "eval_runtime": 32.2846,
+      "eval_samples_per_second": 13.66,
+      "eval_steps_per_second": 3.438,
+      "eval_wer": 0.46573695599903825,
+      "step": 1700
+    },
+    {
+      "epoch": 5.590062111801243,
+      "grad_norm": 4.59730339050293,
+      "learning_rate": 0.0002468096234309623,
+      "loss": 1.1761,
+      "step": 1800
+    },
+    {
+      "epoch": 5.590062111801243,
+      "eval_loss": 0.44803521037101746,
+      "eval_runtime": 31.8849,
+      "eval_samples_per_second": 13.831,
+      "eval_steps_per_second": 3.481,
+      "eval_wer": 0.4674200528973311,
+      "step": 1800
+    },
+    {
+      "epoch": 5.900621118012422,
+      "grad_norm": 2.9201653003692627,
+      "learning_rate": 0.0002436715481171548,
+      "loss": 1.0997,
+      "step": 1900
+    },
+    {
+      "epoch": 5.900621118012422,
+      "eval_loss": 0.4469524919986725,
+      "eval_runtime": 31.9803,
+      "eval_samples_per_second": 13.79,
+      "eval_steps_per_second": 3.471,
+      "eval_wer": 0.4652560711709546,
+      "step": 1900
+    },
+    {
+      "epoch": 6.211180124223603,
+      "grad_norm": 2.5859551429748535,
+      "learning_rate": 0.00024053347280334725,
+      "loss": 1.1203,
+      "step": 2000
+    },
+    {
+      "epoch": 6.211180124223603,
+      "eval_loss": 0.44207584857940674,
+      "eval_runtime": 32.0266,
+      "eval_samples_per_second": 13.77,
+      "eval_steps_per_second": 3.466,
+      "eval_wer": 0.46140899254628515,
+      "step": 2000
+    },
+    {
+      "epoch": 6.521739130434782,
+      "grad_norm": 2.423842191696167,
+      "learning_rate": 0.00023739539748953974,
+      "loss": 1.0749,
+      "step": 2100
+    },
+    {
+      "epoch": 6.521739130434782,
+      "eval_loss": 0.4343813359737396,
+      "eval_runtime": 32.257,
+      "eval_samples_per_second": 13.671,
+      "eval_steps_per_second": 3.441,
+      "eval_wer": 0.4505890839144025,
+      "step": 2100
+    },
+    {
+      "epoch": 6.832298136645963,
+      "grad_norm": 8.447519302368164,
+      "learning_rate": 0.0002342573221757322,
+      "loss": 1.1156,
+      "step": 2200
+    },
+    {
+      "epoch": 6.832298136645963,
+      "eval_loss": 0.43543142080307007,
+      "eval_runtime": 31.9424,
+      "eval_samples_per_second": 13.806,
+      "eval_steps_per_second": 3.475,
+      "eval_wer": 0.45106996874248617,
+      "step": 2200
+    },
+    {
+      "epoch": 7.142857142857143,
+      "grad_norm": 2.308629274368286,
+      "learning_rate": 0.00023111924686192465,
+      "loss": 1.0404,
+      "step": 2300
+    },
+    {
+      "epoch": 7.142857142857143,
+      "eval_loss": 0.43639999628067017,
+      "eval_runtime": 32.0384,
+      "eval_samples_per_second": 13.765,
+      "eval_steps_per_second": 3.465,
+      "eval_wer": 0.45347439288290453,
+      "step": 2300
+    },
+    {
+      "epoch": 7.453416149068323,
+      "grad_norm": 1.2582736015319824,
+      "learning_rate": 0.00022798117154811714,
+      "loss": 1.1081,
+      "step": 2400
+    },
+    {
+      "epoch": 7.453416149068323,
+      "eval_loss": 0.43768569827079773,
+      "eval_runtime": 31.9331,
+      "eval_samples_per_second": 13.81,
+      "eval_steps_per_second": 3.476,
+      "eval_wer": 0.45155085357056984,
+      "step": 2400
+    },
+    {
+      "epoch": 7.763975155279503,
+      "grad_norm": 21.993574142456055,
+      "learning_rate": 0.0002248430962343096,
+      "loss": 1.0535,
+      "step": 2500
+    },
+    {
+      "epoch": 7.763975155279503,
+      "eval_loss": 0.44070157408714294,
+      "eval_runtime": 32.2891,
+      "eval_samples_per_second": 13.658,
+      "eval_steps_per_second": 3.438,
+      "eval_wer": 0.4520317383986535,
+      "step": 2500
+    },
+    {
+      "epoch": 7.763975155279503,
+      "step": 2500,
+      "total_flos": 1.301164200639248e+19,
+      "train_loss": 1.6717456512451172,
+      "train_runtime": 3684.5977,
+      "train_samples_per_second": 20.949,
+      "train_steps_per_second": 2.622
+    }
+  ],
+  "logging_steps": 100,
+  "max_steps": 9660,
+  "num_input_tokens_seen": 0,
+  "num_train_epochs": 30,
+  "save_steps": 400,
+  "stateful_callbacks": {
+    "EarlyStoppingCallback": {
+      "args": {
+        "early_stopping_patience": 4,
+        "early_stopping_threshold": 0.0
+      },
+      "attributes": {
+        "early_stopping_patience_counter": 3
+      }
+    },
+    "TrainerControl": {
+      "args": {
+        "should_epoch_stop": false,
+        "should_evaluate": false,
+        "should_log": false,
+        "should_save": true,
+        "should_training_stop": false
+      },
+      "attributes": {}
+    }
+  },
+  "total_flos": 1.301164200639248e+19,
+  "train_batch_size": 4,
+  "trial_name": null,
+  "trial_params": null
+}