Training in progress, step 100

config.json

@@ -1,5 +1,5 @@
 {
-  "_name_or_path": "vasista22/whisper-hindi-small",
+  "_name_or_path": "qanastek/whisper-small-french-uncased",
   "activation_dropout": 0.0,
   "activation_function": "gelu",
   "apply_spec_augment": false,

model.safetensors

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:96793a88a265c07ef4184ca8fe117b013d58b2b7c85cc356aa07bbb9dcbcf3a6
+oid sha256:3622ab230d796bea00496f6ee443c80ebb2a7d176777ff4817c17d1c4347bb50
 size 966995080

run_cmn.sh

@@ -1,23 +1,23 @@
 python3 run_speech_recognition_seq2seq_streaming.py \
-	--model_name_or_path="vasista22/whisper-hindi-small" \
+	--model_name_or_path="qanastek/whisper-small-french-uncased" \
 	--dataset_name="mozilla-foundation/common_voice_16_0" \
-	--dataset_config_name="hi" \
-	--language="hindi" \
+	--dataset_config_name="fr" \
+	--language="french" \
 	--train_split_name="train+validation" \
 	--eval_split_name="test" \
-	--model_index_name="Whisper Base Bengali" \
-	--max_steps="100" \
+	--model_index_name="Whisper Base French" \
+	--max_steps="1000" \
 	--output_dir="./" \
 	--per_device_train_batch_size="16" \
-	--per_device_eval_batch_size="16" \
+	--per_device_eval_batch_size="32" \
 	--gradient_accumulation_steps=4 \
 	--logging_steps="25" \
-	--learning_rate="1e-6" \
-	--warmup_steps="10" \
+	--learning_rate="5e-7" \
+	--warmup_steps="100" \
 	--evaluation_strategy="steps" \
-	--eval_steps="50" \
+	--eval_steps="100" \
 	--save_strategy="steps" \
-	--save_steps="50" \
+	--save_steps="100" \
 	--generation_max_length="225" \
 	--length_column_name="input_length" \
 	--max_duration_in_seconds="30" \

runs/Jan23_05-57-07_ip-172-31-68-44/events.out.tfevents.1705989446.ip-172-31-68-44.20295.0

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:0486f429675408de46cc06040fcd0b4163ab6cde2d89a5eebc319fd79fdd9b7f
+size 6071

tokenizer.json

@@ -985,7 +985,7 @@
       },
       {
         "SpecialToken": {
-          "id": "<|hi|>",
+          "id": "<|fr|>",
           "type_id": 0
         }
       },
@@ -1023,7 +1023,7 @@
       },
       {
         "SpecialToken": {
-          "id": "<|hi|>",
+          "id": "<|fr|>",
           "type_id": 0
         }
       },
@@ -1068,13 +1068,13 @@
           "<|endoftext|>"
         ]
       },
-      "<|hi|>": {
-        "id": "<|hi|>",
+      "<|fr|>": {
+        "id": "<|fr|>",
         "ids": [
-          50276
+          50265
         ],
         "tokens": [
-          "<|hi|>"
+          "<|fr|>"
         ]
       },
       "<|notimestamps|>": {

training_args.bin

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:982b5193e750d2177a9de84e737d1bca6d1f68d33c76512f2aec4c722e52b389
+oid sha256:c402a950cc7e208f84081823c51369e4011a49eb775a25599d930a6775401c83
 size 4792