Training in progress, step 5500

pytorch_model.bin

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:16d5bdabf704db530a89e8f651eaa5068007117858043a754242f0b36576ecc4
+oid sha256:9c876b7d34fd726488375597a2bce273243c73ee7031a5a915c7ef4fc68bbe6c
 size 1262381549

run_3.sh

@@ -4,12 +4,12 @@ python run_speech_recognition_ctc.py \
       --model_name_or_path="facebook/wav2vec2-xls-r-300m" \
       --output_dir="./" \
       --overwrite_output_dir="False" \
-      --max_steps="5000" \
-      --per_device_train_batch_size="8" \
-      --per_device_eval_batch_size="8" \
-      --gradient_accumulation_steps="4" \
-      --learning_rate="7.5e-5" \
-      --warmup_steps="2000" \
+      --max_steps="6000" \
+      --per_device_train_batch_size="16" \
+      --per_device_eval_batch_size="16" \
+      --gradient_accumulation_steps="2" \
+      --learning_rate="7.5e-6" \
+      --warmup_steps="1000" \
       --evaluation_strategy="steps" \
       --text_column_name="transcription" \
       --save_steps="500" \
@@ -20,10 +20,10 @@ python run_speech_recognition_ctc.py \
       --eval_metrics wer cer \
       --greater_is_better="False" \
       --load_best_model_at_end \
-      --save_total_limit="3" \
-      --mask_time_prob="0.3" \
+      --save_total_limit="10" \
+      --mask_time_prob="0.6" \
       --mask_time_length="10" \
-      --mask_feature_prob="0.1" \
+      --mask_feature_prob="0.2" \
       --fp16 \
       --mask_feature_length="64" \
       --chars_to_ignore , ? . ! - \; \: \" “ % ‘ ” � \

runs/Dec16_20-12-50_129-146-104-29/1671221741.6851091/events.out.tfevents.1671221741.129-146-104-29.144289.1

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:f8a651984c4fb45e3bc756b5f150fcfd878800854f2d3260816244f2d70ec758
+size 5633

runs/Dec16_20-12-50_129-146-104-29/events.out.tfevents.1671221741.129-146-104-29.144289.0

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:56e31caa25b70fc4db7ea2393c8a95a6baaf10daaa434238c98c22cb18ac421a
+size 13384

special_tokens_map.json

@@ -63,6 +63,20 @@
       "rstrip": false,
       "single_word": false
     },
+    {
+      "content": "</s>",
+      "lstrip": false,
+      "normalized": true,
+      "rstrip": false,
+      "single_word": false
+    },
+    {
+      "content": "<s>",
+      "lstrip": false,
+      "normalized": true,
+      "rstrip": false,
+      "single_word": false
+    },
     {
       "content": "</s>",
       "lstrip": false,

training_args.bin

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:7d156a698a73c6b0f8a73ae7839417d2f38b6a755378e7e32d5e1f28079254bf
+oid sha256:4c7c17c12921a7261ed70978c9e767564ee91ada37852a8e0a0ec24576965c09
 size 3451