add checkpoint 135k

config.json

@@ -0,0 +1,85 @@
+{
+  "_name_or_path": "facebook/wav2vec2-large-xlsr-53",
+  "activation_dropout": 0.0,
+  "apply_spec_augment": true,
+  "architectures": [
+    "Wav2Vec2ForCTC"
+  ],
+  "attention_dropout": 0.1,
+  "bos_token_id": 1,
+  "codevector_dim": 768,
+  "contrastive_logits_temperature": 0.1,
+  "conv_bias": true,
+  "conv_dim": [
+    512,
+    512,
+    512,
+    512,
+    512,
+    512,
+    512
+  ],
+  "conv_kernel": [
+    10,
+    3,
+    3,
+    3,
+    3,
+    2,
+    2
+  ],
+  "conv_stride": [
+    5,
+    2,
+    2,
+    2,
+    2,
+    2,
+    2
+  ],
+  "ctc_loss_reduction": "mean",
+  "ctc_zero_infinity": false,
+  "diversity_loss_weight": 0.1,
+  "do_stable_layer_norm": true,
+  "eos_token_id": 2,
+  "feat_extract_activation": "gelu",
+  "feat_extract_dropout": 0.0,
+  "feat_extract_norm": "layer",
+  "feat_proj_dropout": 0.0,
+  "feat_quantizer_dropout": 0.0,
+  "final_dropout": 0.0,
+  "gradient_checkpointing": true,
+  "hidden_act": "gelu",
+  "hidden_dropout": 0.1,
+  "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_prob": 0.0,
+  "mask_time_length": 10,
+  "mask_time_min_space": 1,
+  "mask_time_other": 0.0,
+  "mask_time_prob": 0.05,
+  "mask_time_selection": "static",
+  "model_type": "wav2vec2",
+  "num_attention_heads": 16,
+  "num_codevector_groups": 2,
+  "num_codevectors_per_group": 320,
+  "num_conv_pos_embedding_groups": 16,
+  "num_conv_pos_embeddings": 128,
+  "num_feat_extract_layers": 7,
+  "num_hidden_layers": 24,
+  "num_negatives": 100,
+  "pad_token_id": 69,
+  "proj_codevector_dim": 768,
+  "torch_dtype": "float32",
+  "transformers_version": "4.9.1",
+  "vocab_size": 70
+}

preprocessor_config.json

@@ -0,0 +1,9 @@
+{
+  "do_normalize": true,
+  "feature_extractor_type": "Wav2Vec2FeatureExtractor",
+  "feature_size": 1,
+  "padding_side": "right",
+  "padding_value": 0.0,
+  "return_attention_mask": false,
+  "sampling_rate": 16000
+}

rng_state.pth

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:040892c8367ed2663b43590614fcaee20e683cc064af55f65e3baf94db0473d8
+size 15587

scaler.pt

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:eaad0a550aad796c00c2715fb3dd07c74e5ca1520507a15a505e4faf0123239f
+size 559

scheduler.pt

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:bb118fee7b3cd82205b90d8e4ffc01eee7dfb1f3f32f5210cf6b36e69150bc49
+size 623

trainer_state.json

@@ -0,0 +1,2041 @@
+{
+  "best_metric": 1.0,
+  "best_model_checkpoint": "../data/wav2vec2-large-xlsr-53-thai/checkpoint-1000",
+  "epoch": 99.77827050997783,
+  "global_step": 135000,
+  "is_hyper_param_search": false,
+  "is_local_process_zero": true,
+  "is_world_process_zero": true,
+  "log_history": [
+    {
+      "epoch": 0.74,
+      "learning_rate": 9.970000000000001e-05,
+      "loss": 7.1725,
+      "step": 1000
+    },
+    {
+      "epoch": 0.74,
+      "eval_loss": 3.039940595626831,
+      "eval_runtime": 185.7874,
+      "eval_samples_per_second": 16.293,
+      "eval_steps_per_second": 0.511,
+      "eval_wer": 1.0,
+      "step": 1000
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 9.925763216679077e-05,
+      "loss": 1.0788,
+      "step": 2000
+    },
+    {
+      "epoch": 1.48,
+      "eval_loss": 0.6130112409591675,
+      "eval_runtime": 176.9194,
+      "eval_samples_per_second": 17.109,
+      "eval_steps_per_second": 0.537,
+      "eval_wer": 0.6820256330103157,
+      "step": 2000
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 9.851303052866717e-05,
+      "loss": 0.6339,
+      "step": 3000
+    },
+    {
+      "epoch": 2.22,
+      "eval_loss": 0.49624550342559814,
+      "eval_runtime": 175.6715,
+      "eval_samples_per_second": 17.231,
+      "eval_steps_per_second": 0.541,
+      "eval_wer": 0.6168802750859643,
+      "step": 3000
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 9.776842889054357e-05,
+      "loss": 0.5295,
+      "step": 4000
+    },
+    {
+      "epoch": 2.96,
+      "eval_loss": 0.4389719069004059,
+      "eval_runtime": 176.6072,
+      "eval_samples_per_second": 17.14,
+      "eval_steps_per_second": 0.538,
+      "eval_wer": 0.577117849327915,
+      "step": 4000
+    },
+    {
+      "epoch": 3.7,
+      "learning_rate": 9.702382725241996e-05,
+      "loss": 0.4616,
+      "step": 5000
+    },
+    {
+      "epoch": 3.7,
+      "eval_loss": 0.4208119809627533,
+      "eval_runtime": 196.7481,
+      "eval_samples_per_second": 15.385,
+      "eval_steps_per_second": 0.483,
+      "eval_wer": 0.5618005626758362,
+      "step": 5000
+    },
+    {
+      "epoch": 4.43,
+      "learning_rate": 9.627922561429635e-05,
+      "loss": 0.4206,
+      "step": 6000
+    },
+    {
+      "epoch": 4.43,
+      "eval_loss": 0.4009915888309479,
+      "eval_runtime": 180.4803,
+      "eval_samples_per_second": 16.772,
+      "eval_steps_per_second": 0.526,
+      "eval_wer": 0.5411691153485464,
+      "step": 6000
+    },
+    {
+      "epoch": 5.17,
+      "learning_rate": 9.553462397617276e-05,
+      "loss": 0.3908,
+      "step": 7000
+    },
+    {
+      "epoch": 5.17,
+      "eval_loss": 0.399889200925827,
+      "eval_runtime": 189.7255,
+      "eval_samples_per_second": 15.955,
+      "eval_steps_per_second": 0.501,
+      "eval_wer": 0.5384807752422632,
+      "step": 7000
+    },
+    {
+      "epoch": 5.91,
+      "learning_rate": 9.479002233804914e-05,
+      "loss": 0.3659,
+      "step": 8000
+    },
+    {
+      "epoch": 5.91,
+      "eval_loss": 0.3719017505645752,
+      "eval_runtime": 184.5994,
+      "eval_samples_per_second": 16.398,
+      "eval_steps_per_second": 0.515,
+      "eval_wer": 0.5141606752110034,
+      "step": 8000
+    },
+    {
+      "epoch": 6.65,
+      "learning_rate": 9.404542069992555e-05,
+      "loss": 0.338,
+      "step": 9000
+    },
+    {
+      "epoch": 6.65,
+      "eval_loss": 0.35533592104911804,
+      "eval_runtime": 188.5325,
+      "eval_samples_per_second": 16.056,
+      "eval_steps_per_second": 0.504,
+      "eval_wer": 0.5101594248202563,
+      "step": 9000
+    },
+    {
+      "epoch": 7.39,
+      "learning_rate": 9.330156366344007e-05,
+      "loss": 0.3222,
+      "step": 10000
+    },
+    {
+      "epoch": 7.39,
+      "eval_loss": 0.40722423791885376,
+      "eval_runtime": 195.2723,
+      "eval_samples_per_second": 15.501,
+      "eval_steps_per_second": 0.487,
+      "eval_wer": 0.49309159112222567,
+      "step": 10000
+    },
+    {
+      "epoch": 8.13,
+      "learning_rate": 9.255696202531646e-05,
+      "loss": 0.31,
+      "step": 11000
+    },
+    {
+      "epoch": 8.13,
+      "eval_loss": 0.37148234248161316,
+      "eval_runtime": 182.6314,
+      "eval_samples_per_second": 16.574,
+      "eval_steps_per_second": 0.52,
+      "eval_wer": 0.48715223507346045,
+      "step": 11000
+    },
+    {
+      "epoch": 8.87,
+      "learning_rate": 9.181310498883098e-05,
+      "loss": 0.2901,
+      "step": 12000
+    },
+    {
+      "epoch": 8.87,
+      "eval_loss": 0.365583598613739,
+      "eval_runtime": 175.3968,
+      "eval_samples_per_second": 17.258,
+      "eval_steps_per_second": 0.542,
+      "eval_wer": 0.4882775867458581,
+      "step": 12000
+    },
+    {
+      "epoch": 9.61,
+      "learning_rate": 9.106850335070738e-05,
+      "loss": 0.276,
+      "step": 13000
+    },
+    {
+      "epoch": 9.61,
+      "eval_loss": 0.37835174798965454,
+      "eval_runtime": 175.9962,
+      "eval_samples_per_second": 17.199,
+      "eval_steps_per_second": 0.54,
+      "eval_wer": 0.4897155361050328,
+      "step": 13000
+    },
+    {
+      "epoch": 10.35,
+      "learning_rate": 9.032539091586003e-05,
+      "loss": 0.2628,
+      "step": 14000
+    },
+    {
+      "epoch": 10.35,
+      "eval_loss": 0.4127887189388275,
+      "eval_runtime": 180.0512,
+      "eval_samples_per_second": 16.812,
+      "eval_steps_per_second": 0.528,
+      "eval_wer": 0.48871522350734603,
+      "step": 14000
+    },
+    {
+      "epoch": 11.09,
+      "learning_rate": 8.958078927773641e-05,
+      "loss": 0.256,
+      "step": 15000
+    },
+    {
+      "epoch": 11.09,
+      "eval_loss": 0.3795180022716522,
+      "eval_runtime": 174.1363,
+      "eval_samples_per_second": 17.383,
+      "eval_steps_per_second": 0.546,
+      "eval_wer": 0.46939668646452015,
+      "step": 15000
+    },
+    {
+      "epoch": 11.83,
+      "learning_rate": 8.88361876396128e-05,
+      "loss": 0.2455,
+      "step": 16000
+    },
+    {
+      "epoch": 11.83,
+      "eval_loss": 0.3748489320278168,
+      "eval_runtime": 174.8629,
+      "eval_samples_per_second": 17.311,
+      "eval_steps_per_second": 0.543,
+      "eval_wer": 0.46970928415129726,
+      "step": 16000
+    },
+    {
+      "epoch": 12.56,
+      "learning_rate": 8.809233060312732e-05,
+      "loss": 0.2377,
+      "step": 17000
+    },
+    {
+      "epoch": 12.56,
+      "eval_loss": 0.4371410012245178,
+      "eval_runtime": 173.1913,
+      "eval_samples_per_second": 17.478,
+      "eval_steps_per_second": 0.549,
+      "eval_wer": 0.4650203188496405,
+      "step": 17000
+    },
+    {
+      "epoch": 13.3,
+      "learning_rate": 8.734772896500373e-05,
+      "loss": 0.2279,
+      "step": 18000
+    },
+    {
+      "epoch": 13.3,
+      "eval_loss": 0.38019847869873047,
+      "eval_runtime": 177.0444,
+      "eval_samples_per_second": 17.097,
+      "eval_steps_per_second": 0.537,
+      "eval_wer": 0.45339168490153176,
+      "step": 18000
+    },
+    {
+      "epoch": 14.04,
+      "learning_rate": 8.660387192851825e-05,
+      "loss": 0.2247,
+      "step": 19000
+    },
+    {
+      "epoch": 14.04,
+      "eval_loss": 0.3687661588191986,
+      "eval_runtime": 197.8618,
+      "eval_samples_per_second": 15.299,
+      "eval_steps_per_second": 0.48,
+      "eval_wer": 0.4567677399187246,
+      "step": 19000
+    },
+    {
+      "epoch": 14.78,
+      "learning_rate": 8.585927029039465e-05,
+      "loss": 0.2139,
+      "step": 20000
+    },
+    {
+      "epoch": 14.78,
+      "eval_loss": 0.3616667091846466,
+      "eval_runtime": 177.8968,
+      "eval_samples_per_second": 17.015,
+      "eval_steps_per_second": 0.534,
+      "eval_wer": 0.446639574867146,
+      "step": 20000
+    },
+    {
+      "epoch": 15.52,
+      "learning_rate": 8.511541325390916e-05,
+      "loss": 0.2072,
+      "step": 21000
+    },
+    {
+      "epoch": 15.52,
+      "eval_loss": 0.4171736538410187,
+      "eval_runtime": 171.8647,
+      "eval_samples_per_second": 17.613,
+      "eval_steps_per_second": 0.553,
+      "eval_wer": 0.45883088465145355,
+      "step": 21000
+    },
+    {
+      "epoch": 16.26,
+      "learning_rate": 8.437081161578556e-05,
+      "loss": 0.1999,
+      "step": 22000
+    },
+    {
+      "epoch": 16.26,
+      "eval_loss": 0.7755644917488098,
+      "eval_runtime": 171.1132,
+      "eval_samples_per_second": 17.69,
+      "eval_steps_per_second": 0.555,
+      "eval_wer": 0.45126602063144733,
+      "step": 22000
+    },
+    {
+      "epoch": 17.0,
+      "learning_rate": 8.362695457930007e-05,
+      "loss": 0.1958,
+      "step": 23000
+    },
+    {
+      "epoch": 17.0,
+      "eval_loss": 0.41745948791503906,
+      "eval_runtime": 172.9131,
+      "eval_samples_per_second": 17.506,
+      "eval_steps_per_second": 0.549,
+      "eval_wer": 0.44382619568615195,
+      "step": 23000
+    },
+    {
+      "epoch": 17.74,
+      "learning_rate": 8.288235294117647e-05,
+      "loss": 0.1866,
+      "step": 24000
+    },
+    {
++