Training in progress, step 500

config.json

@@ -42,7 +42,7 @@
     2
   ],
   "ctc_loss_reduction": "mean",
-  "ctc_zero_infinity": false,
+  "ctc_zero_infinity": true,
   "diversity_loss_weight": 0.1,
   "do_stable_layer_norm": true,
   "eos_token_id": 2,
@@ -58,7 +58,7 @@
   "initializer_range": 0.02,
   "intermediate_size": 4096,
   "layer_norm_eps": 1e-05,
-  "layerdrop": 0.0,
+  "layerdrop": 0.1,
   "mask_feature_length": 64,
   "mask_feature_min_masks": 0,
   "mask_feature_prob": 0.25,

pytorch_model.bin

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:803e310024afc447cc9af76806d965bd3dfe885afbcc9f5a8a9469feb7102c4f
+oid sha256:801ed6a8fee74d72d6fc0685e04cd256f1e03e216b971c82295eba8597bb5a44
 size 1262067185

run.sh

@@ -1,7 +1,7 @@
 python run_speech_recognition_ctc.py \
         --dataset_name="NbAiLab/NPSC" \
         --model_name_or_path="facebook/wav2vec2-xls-r-300m" \
-        --hub_model_id="NbAiLab/wav2vec2-xlsr-300M-NPSC-OH" \
+	--hub_model_id="NbAiLab/wav2vec2-xlsr-300M-NPSC-OH" \
 	--dataset_config_name="16K_mp3" \
         --output_dir="./" \
         --overwrite_output_dir \
@@ -17,7 +17,7 @@ python run_speech_recognition_ctc.py \
         --save_steps="500" \
         --eval_steps="500" \
         --logging_steps="100" \
-        --layerdrop="0.0" \
+        --layerdrop="0.1" \
         --activation_dropout="0.1" \
         --save_total_limit="3" \
         --freeze_feature_encoder \

run_speech_recognition_ctc.py

@@ -400,8 +400,10 @@ def main():
         return re.search("nb-no", entry["sentence_language_code"], flags=re.IGNORECASE)
 
     def filter_tooshort(entry):
-        print(f"The audio sample ({entry["audio"]["path"]}) is too small, and has been omitted. "
-        return not len(entry["audio"]["array"] // 320 >= len(entry["text"])
+        if len(entry["audio"]["array"]) // 320 < len(entry["text"]):
+            print(f"The audio sample ({entry['audio']['path']}) is too small, and has been omitted. ")
+            return False
+        return True
 
     def map_dataset(entry):
         batch = {"text": entry["text"].lower()}
@@ -570,6 +572,7 @@ def main():
             "gradient_checkpointing": training_args.gradient_checkpointing,
             "layerdrop": model_args.layerdrop,
             "ctc_loss_reduction": model_args.ctc_loss_reduction,
+            "ctc_zero_infinity": True,
             "pad_token_id": tokenizer.pad_token_id,
             "vocab_size": len(tokenizer),
             "activation_dropout": model_args.activation_dropout,

runs/Feb01_12-24-21_job-c93f32d8-97c5-48e7-b5ec-c6c950f627ba/1643718355.8126762/events.out.tfevents.1643718355.job-c93f32d8-97c5-48e7-b5ec-c6c950f627ba.2341977.1

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:88d8214f159fbe7516815824430a8b5063ea85a2c81ec9b3aa643edb48f424cf
+size 4783

runs/Feb01_12-24-21_job-c93f32d8-97c5-48e7-b5ec-c6c950f627ba/events.out.tfevents.1643718355.job-c93f32d8-97c5-48e7-b5ec-c6c950f627ba.2341977.0

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:c958c822ada84b1247c0379441b9259329a8e242947ae3b15492c8bdbf124b2e
+size 5815

special_tokens_map.json

@@ -1 +1 @@
-{"bos_token": "<s>", "eos_token": "</s>", "unk_token": "[UNK]", "pad_token": "[PAD]", "additional_special_tokens": [{"content": "<s>", "single_word": false, "lstrip": false, "rstrip": false, "normalized": true}, {"content": "</s>", "single_word": false, "lstrip": false, "rstrip": false, "normalized": true}, {"content": "<s>", "single_word": false, "lstrip": false, "rstrip": false, "normalized": true}, {"content": "</s>", "single_word": false, "lstrip": false, "rstrip": false, "normalized": true}, {"content": "<s>", "single_word": false, "lstrip": false, "rstrip": false, "normalized": true}, {"content": "</s>", "single_word": false, "lstrip": false, "rstrip": false, "normalized": true}, {"content": "<s>", "single_word": false, "lstrip": false, "rstrip": false, "normalized": true}, {"content": "</s>", "single_word": false, "lstrip": false, "rstrip": false, "normalized": true}, {"content": "<s>", "single_word": false, "lstrip": false, "rstrip": false, "normalized": true}, {"content": "</s>", "single_word": false, "lstrip": false, "rstrip": false, "normalized": true}, {"content": "<s>", "single_word": false, "lstrip": false, "rstrip": false, "normalized": true}, {"content": "</s>", "single_word": false, "lstrip": false, "rstrip": false, "normalized": true}]}
+{"bos_token": "<s>", "eos_token": "</s>", "unk_token": "[UNK]", "pad_token": "[PAD]", "additional_special_tokens": [{"content": "<s>", "single_word": false, "lstrip": false, "rstrip": false, "normalized": true}, {"content": "</s>", "single_word": false, "lstrip": false, "rstrip": false, "normalized": true}]}

training_args.bin

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:14b87517d531e6c25e7ad265bb8803b452a69564e9db407d395cede3f8b33a49
+oid sha256:642fc6041127b9c77f68ff427a0324d8a66f227388841dc1c5a1070b6b9fabb9
 size 3055