Upload 8 files

config.json

@@ -1,6 +1,6 @@
 {
-  "_name_or_path": "elgeish/wav2vec2-large-xlsr-53-arabic",
-  "activation_dropout": 0.1,
+  "_name_or_path": "facebook/wav2vec2-large-xlsr-53",
+  "activation_dropout": 0.0,
   "adapter_kernel_size": 3,
   "adapter_stride": 2,
   "add_adapter": false,
@@ -11,7 +11,7 @@
   "attention_dropout": 0.1,
   "bos_token_id": 1,
   "classifier_proj_size": 256,
-  "codevector_dim": 256,
+  "codevector_dim": 768,
   "contrastive_logits_temperature": 0.1,
   "conv_bias": true,
   "conv_dim": [
@@ -41,7 +41,7 @@
     2,
     2
   ],
-  "ctc_loss_reduction": "sum",
+  "ctc_loss_reduction": "mean",
   "ctc_zero_infinity": false,
   "diversity_loss_weight": 0.1,
   "do_stable_layer_norm": true,
@@ -49,24 +49,30 @@
   "feat_extract_activation": "gelu",
   "feat_extract_dropout": 0.0,
   "feat_extract_norm": "layer",
-  "feat_proj_dropout": 0.1,
+  "feat_proj_dropout": 0.0,
   "feat_quantizer_dropout": 0.0,
-  "final_dropout": 0.1,
-  "gradient_checkpointing": false,
+  "final_dropout": 0.0,
   "hidden_act": "gelu",
-  "hidden_dropout": 0.1,
-  "hidden_dropout_prob": 0.1,
+  "hidden_dropout": 0.05,
   "hidden_size": 1024,
   "initializer_range": 0.02,
   "intermediate_size": 4096,
   "layer_norm_eps": 1e-05,
   "layerdrop": 0.1,
+  "mask_channel_length": 10,
+  "mask_channel_min_space": 1,
+  "mask_channel_other": 0.0,
+  "mask_channel_prob": 0.0,
+  "mask_channel_selection": "static",
   "mask_feature_length": 10,
   "mask_feature_min_masks": 0,
   "mask_feature_prob": 0.0,
   "mask_time_length": 10,
   "mask_time_min_masks": 2,
+  "mask_time_min_space": 1,
+  "mask_time_other": 0.0,
   "mask_time_prob": 0.05,
+  "mask_time_selection": "static",
   "model_type": "wav2vec2",
   "num_adapter_layers": 3,
   "num_attention_heads": 16,
@@ -78,8 +84,8 @@
   "num_hidden_layers": 24,
   "num_negatives": 100,
   "output_hidden_size": 1024,
-  "pad_token_id": 0,
-  "proj_codevector_dim": 256,
+  "pad_token_id": 36,
+  "proj_codevector_dim": 768,
   "tdnn_dilation": [
     1,
     2,
@@ -104,6 +110,6 @@
   "torch_dtype": "float32",
   "transformers_version": "4.25.1",
   "use_weighted_layer_sum": false,
-  "vocab_size": 56,
+  "vocab_size": 37,
   "xvector_output_dim": 512
 }

pytorch_model.bin

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:f20c1031dc99ad9df87d48e48e73eefece59ed3738594c7388e2dcad81fadbf0
-size 1262128305
+oid sha256:fe8505ec4ad5178b4bc9516c0864ec6baef9af7146a654d09dab5d7bfdf089a0
+size 1262050417

special_tokens_map.json

@@ -1 +1,6 @@
-{"bos_token": "<s>", "eos_token": "</s>", "unk_token": "[UNK]", "pad_token": "[PAD]"}
+{
+  "bos_token": "<s>",
+  "eos_token": "</s>",
+  "pad_token": "[PAD]",
+  "unk_token": "[UNK]"
+}

trainer_state.json

@@ -0,0 +1,256 @@
+{
+  "best_metric": null,
+  "best_model_checkpoint": null,
+  "epoch": 27.0042194092827,
+  "global_step": 6400,
+  "is_hyper_param_search": false,
+  "is_local_process_zero": true,
+  "is_world_process_zero": true,
+  "log_history": [
+    {
+      "epoch": 1.69,
+      "learning_rate": 7.920000000000001e-05,
+      "loss": 9.7705,
+      "step": 400
+    },
+    {
+      "epoch": 1.69,
+      "eval_loss": 3.045895576477051,
+      "eval_runtime": 105.6261,
+      "eval_samples_per_second": 16.047,
+      "eval_steps_per_second": 2.007,
+      "eval_wer": 1.0,
+      "step": 400
+    },
+    {
+      "epoch": 3.38,
+      "learning_rate": 9.517601043024771e-05,
+      "loss": 2.7947,
+      "step": 800
+    },
+    {
+      "epoch": 3.38,
+      "eval_loss": 1.6579201221466064,
+      "eval_runtime": 102.9312,
+      "eval_samples_per_second": 16.467,
+      "eval_steps_per_second": 2.06,
+      "eval_wer": 1.0785922937821673,
+      "step": 800
+    },
+    {
+      "epoch": 5.06,
+      "learning_rate": 8.865710560625816e-05,
+      "loss": 1.0354,
+      "step": 1200
+    },
+    {
+      "epoch": 5.06,
+      "eval_loss": 0.6200010776519775,
+      "eval_runtime": 102.9566,
+      "eval_samples_per_second": 16.463,
+      "eval_steps_per_second": 2.059,
+      "eval_wer": 0.7493392683266101,
+      "step": 1200
+    },
+    {
+      "epoch": 6.75,
+      "learning_rate": 8.213820078226858e-05,
+      "loss": 0.5782,
+      "step": 1600
+    },
+    {
+      "epoch": 6.75,
+      "eval_loss": 0.5652060508728027,
+      "eval_runtime": 104.7594,
+      "eval_samples_per_second": 16.18,
+      "eval_steps_per_second": 2.024,
+      "eval_wer": 0.7191542634580609,
+      "step": 1600
+    },
+    {
+      "epoch": 8.44,
+      "learning_rate": 7.561929595827901e-05,
+      "loss": 0.4526,
+      "step": 2000
+    },
+    {
+      "epoch": 8.44,
+      "eval_loss": 0.5032265782356262,
+      "eval_runtime": 118.2197,
+      "eval_samples_per_second": 14.338,
+      "eval_steps_per_second": 1.793,
+      "eval_wer": 0.6482125469467241,
+      "step": 2000
+    },
+    {
+      "epoch": 10.13,
+      "learning_rate": 6.910039113428944e-05,
+      "loss": 0.3766,
+      "step": 2400
+    },
+    {
+      "epoch": 10.13,
+      "eval_loss": 0.4892590045928955,
+      "eval_runtime": 106.8164,
+      "eval_samples_per_second": 15.868,
+      "eval_steps_per_second": 1.985,
+      "eval_wer": 0.6219223814160523,
+      "step": 2400
+    },
+    {
+      "epoch": 11.81,
+      "learning_rate": 6.258148631029988e-05,
+      "loss": 0.3112,
+      "step": 2800
+    },
+    {
+      "epoch": 11.81,
+      "eval_loss": 0.46891558170318604,
+      "eval_runtime": 105.9027,
+      "eval_samples_per_second": 16.005,
+      "eval_steps_per_second": 2.002,
+      "eval_wer": 0.6139936013353735,
+      "step": 2800
+    },
+    {
+      "epoch": 13.5,
+      "learning_rate": 5.606258148631031e-05,
+      "loss": 0.2825,
+      "step": 3200
+    },
+    {
+      "epoch": 13.5,
+      "eval_loss": 0.48954522609710693,
+      "eval_runtime": 108.9558,
+      "eval_samples_per_second": 15.557,
+      "eval_steps_per_second": 1.946,
+      "eval_wer": 0.6003616636528029,
+      "step": 3200
+    },
+    {
+      "epoch": 15.19,
+      "learning_rate": 4.954367666232073e-05,
+      "loss": 0.253,
+      "step": 3600
+    },
+    {
+      "epoch": 15.19,
+      "eval_loss": 0.494277685880661,
+      "eval_runtime": 107.9999,
+      "eval_samples_per_second": 15.694,
+      "eval_steps_per_second": 1.963,
+      "eval_wer": 0.5914591737376548,
+      "step": 3600
+    },
+    {
+      "epoch": 16.88,
+      "learning_rate": 4.3024771838331163e-05,
+      "loss": 0.2359,
+      "step": 4000
+    },
+    {
+      "epoch": 16.88,
+      "eval_loss": 0.47655120491981506,
+      "eval_runtime": 110.482,
+      "eval_samples_per_second": 15.342,
+      "eval_steps_per_second": 1.919,
+      "eval_wer": 0.5760189177910697,
+      "step": 4000
+    },
+    {
+      "epoch": 18.57,
+      "learning_rate": 3.650586701434159e-05,
+      "loss": 0.2086,
+      "step": 4400
+    },
+    {
+      "epoch": 18.57,
+      "eval_loss": 0.48324596881866455,
+      "eval_runtime": 105.3701,
+      "eval_samples_per_second": 16.086,
+      "eval_steps_per_second": 2.012,
+      "eval_wer": 0.57768813465016,
+      "step": 4400
+    },
+    {
+      "epoch": 20.25,
+      "learning_rate": 2.9986962190352026e-05,
+      "loss": 0.1987,
+      "step": 4800
+    },
+    {
+      "epoch": 20.25,
+      "eval_loss": 0.4831419885158539,
+      "eval_runtime": 102.4502,
+      "eval_samples_per_second": 16.545,
+      "eval_steps_per_second": 2.069,
+      "eval_wer": 0.5722631798581166,
+      "step": 4800
+    },
+    {
+      "epoch": 21.94,
+      "learning_rate": 2.3468057366362453e-05,
+      "loss": 0.1812,
+      "step": 5200
+    },
+    {
+      "epoch": 21.94,
+      "eval_loss": 0.47938045859336853,
+      "eval_runtime": 110.2812,
+      "eval_samples_per_second": 15.37,
+      "eval_steps_per_second": 1.922,
+      "eval_wer": 0.5669773264709974,
+      "step": 5200
+    },
+    {
+      "epoch": 23.63,
+      "learning_rate": 1.694915254237288e-05,
+      "loss": 0.1759,
+      "step": 5600
+    },
+    {
+      "epoch": 23.63,
+      "eval_loss": 0.4787568747997284,
+      "eval_runtime": 111.7955,
+      "eval_samples_per_second": 15.162,
+      "eval_steps_per_second": 1.896,
+      "eval_wer": 0.564473501182362,
+      "step": 5600
+    },
+    {
+      "epoch": 25.32,
+      "learning_rate": 1.0430247718383312e-05,
+      "loss": 0.1695,
+      "step": 6000
+    },
+    {
+      "epoch": 25.32,
+      "eval_loss": 0.471430242061615,
+      "eval_runtime": 115.4715,
+      "eval_samples_per_second": 14.679,
+      "eval_steps_per_second": 1.836,
+      "eval_wer": 0.5600222562247879,
+      "step": 6000
+    },
+    {
+      "epoch": 27.0,
+      "learning_rate": 3.9113428943937425e-06,
+      "loss": 0.1595,
+      "step": 6400
+    },
+    {
+      "epoch": 27.0,
+      "eval_loss": 0.4793846905231476,
+      "eval_runtime": 107.2446,
+      "eval_samples_per_second": 15.805,
+      "eval_steps_per_second": 1.977,
+      "eval_wer": 0.5586312421755459,
+      "step": 6400
+    }
+  ],
+  "max_steps": 6636,
+  "num_train_epochs": 28,
+  "total_flos": 1.1173889195953228e+19,
+  "trial_name": null,
+  "trial_params": null
+}

training_args.bin

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:a1678b64c373bd5f8e6c6a9d4921225ba738441c207bd88ba3a3cfa539732923
+oid sha256:3f0844105c9fb573fc3ae8c7bbcd2cb8733a3148021a25c4d225869c0062191c
 size 3439