Hugging Face's logo

Scrya
/
whisper-medium-ms

Automatic Speech Recognition
Transformers
PyTorch
TensorBoard
Malay
whisper
whisper-event
Generated from Trainer
Eval Results
Inference Endpoints
Model card Files Metrics Community
Scrya commited on
Commit
c24737a
•
1 Parent(s): ae31109

Training in progress, step 2000

Browse files
Files changed (4) hide show
  1. .ipynb_checkpoints/fine-tune-whisper-non-streaming-fleurs-ms-checkpoint.ipynb +739 -3
  2. fine-tune-whisper-non-streaming-fleurs-ms.ipynb +16 -10
  3. pytorch_model.bin +1 -1
  4. runs/Dec16_12-04-12_129-213-26-202/events.out.tfevents.1671192266.129-213-26-202.127891.0 +2 -2
.ipynb_checkpoints/fine-tune-whisper-non-streaming-fleurs-ms-checkpoint.ipynb CHANGED
@@ -2711,7 +2711,8 @@
2711
  "execution_count": null,
2712
  "id": "ee8b7b8e-1c9a-4d77-9137-1778a629e6de",
2713
  "metadata": {
2714
- "id": "ee8b7b8e-1c9a-4d77-9137-1778a629e6de"
 
2715
  },
2716
  "outputs": [
2717
  {
@@ -2737,8 +2738,8 @@
2737
  "\n",
2738
  " <div>\n",
2739
  " \n",
2740
- " <progress value='9' max='5000' style='width:300px; height:20px; vertical-align: middle;'></progress>\n",
2741
- " [ 9/5000 00:47 < 9:19:01, 0.15 it/s, Epoch 0.09/54]\n",
2742
  " </div>\n",
2743
  " <table border=\"1\" class=\"dataframe\">\n",
2744
  " <thead>\n",
@@ -2746,9 +2747,16 @@
2746
  " <th>Step</th>\n",
2747
  " <th>Training Loss</th>\n",
2748
  " <th>Validation Loss</th>\n",
 
2749
  " </tr>\n",
2750
  " </thead>\n",
2751
  " <tbody>\n",
 
 
 
 
 
 
2752
  " </tbody>\n",
2753
  "</table><p>"
2754
  ],
@@ -2758,6 +2766,734 @@
2758
  },
2759
  "metadata": {},
2760
  "output_type": "display_data"
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
2761
  }
2762
  ],
2763
  "source": [
 
2711
  "execution_count": null,
2712
  "id": "ee8b7b8e-1c9a-4d77-9137-1778a629e6de",
2713
  "metadata": {
2714
+ "id": "ee8b7b8e-1c9a-4d77-9137-1778a629e6de",
2715
+ "scrolled": false
2716
  },
2717
  "outputs": [
2718
  {
 
2738
  "\n",
2739
  " <div>\n",
2740
  " \n",
2741
+ " <progress value='1014' max='5000' style='width:300px; height:20px; vertical-align: middle;'></progress>\n",
2742
+ " [1014/5000 2:09:20 < 8:29:26, 0.13 it/s, Epoch 10.78/54]\n",
2743
  " </div>\n",
2744
  " <table border=\"1\" class=\"dataframe\">\n",
2745
  " <thead>\n",
 
2747
  " <th>Step</th>\n",
2748
  " <th>Training Loss</th>\n",
2749
  " <th>Validation Loss</th>\n",
2750
+ " <th>Wer</th>\n",
2751
  " </tr>\n",
2752
  " </thead>\n",
2753
  " <tbody>\n",
2754
+ " <tr>\n",
2755
+ " <td>1000</td>\n",
2756
+ " <td>0.002400</td>\n",
2757
+ " <td>0.243783</td>\n",
2758
+ " <td>10.344360</td>\n",
2759
+ " </tr>\n",
2760
  " </tbody>\n",
2761
  "</table><p>"
2762
  ],
 
2766
  },
2767
  "metadata": {},
2768
  "output_type": "display_data"
2769
+ },
2770
+ {
2771
+ "name": "stderr",
2772
+ "output_type": "stream",
2773
+ "text": [
2774
+ "The following columns in the evaluation set don't have a corresponding argument in `WhisperForConditionalGeneration.forward` and have been ignored: input_length. If input_length are not expected by `WhisperForConditionalGeneration.forward`, you can safely ignore this message.\n",
2775
+ "***** Running Evaluation *****\n",
2776
+ " Num examples = 749\n",
2777
+ " Batch size = 16\n",
2778
+ "Generate config GenerationConfig {\n",
2779
+ " \"begin_suppress_tokens\": [\n",
2780
+ " 220,\n",
2781
+ " 50257\n",
2782
+ " ],\n",
2783
+ " \"bos_token_id\": 50257,\n",
2784
+ " \"decoder_start_token_id\": 50258,\n",
2785
+ " \"eos_token_id\": 50257,\n",
2786
+ " \"max_length\": 448,\n",
2787
+ " \"pad_token_id\": 50257,\n",
2788
+ " \"suppress_tokens\": [],\n",
2789
+ " \"transformers_version\": \"4.26.0.dev0\",\n",
2790
+ " \"use_cache\": false\n",
2791
+ "}\n",
2792
+ "\n",
2793
+ "/home/ubuntu/hf_env/lib/python3.8/site-packages/transformers/generation/utils.py:1134: UserWarning: You have modified the pretrained model configuration to control generation. This is a deprecated strategy to control generation and will be removed soon, in a future version. Please use a generation configuration file (see https://huggingface.co/docs/transformers/main_classes/text_generation)\n",
2794
+ " warnings.warn(\n",
2795
+ "Generate config GenerationConfig {\n",
2796
+ " \"begin_suppress_tokens\": [\n",
2797
+ " 220,\n",
2798
+ " 50257\n",
2799
+ " ],\n",
2800
+ " \"bos_token_id\": 50257,\n",
2801
+ " \"decoder_start_token_id\": 50258,\n",
2802
+ " \"eos_token_id\": 50257,\n",
2803
+ " \"max_length\": 448,\n",
2804
+ " \"pad_token_id\": 50257,\n",
2805
+ " \"suppress_tokens\": [],\n",
2806
+ " \"transformers_version\": \"4.26.0.dev0\",\n",
2807
+ " \"use_cache\": false\n",
2808
+ "}\n",
2809
+ "\n",
2810
+ "Generate config GenerationConfig {\n",
2811
+ " \"begin_suppress_tokens\": [\n",
2812
+ " 220,\n",
2813
+ " 50257\n",
2814
+ " ],\n",
2815
+ " \"bos_token_id\": 50257,\n",
2816
+ " \"decoder_start_token_id\": 50258,\n",
2817
+ " \"eos_token_id\": 50257,\n",
2818
+ " \"max_length\": 448,\n",
2819
+ " \"pad_token_id\": 50257,\n",
2820
+ " \"suppress_tokens\": [],\n",
2821
+ " \"transformers_version\": \"4.26.0.dev0\",\n",
2822
+ " \"use_cache\": false\n",
2823
+ "}\n",
2824
+ "\n",
2825
+ "Generate config GenerationConfig {\n",
2826
+ " \"begin_suppress_tokens\": [\n",
2827
+ " 220,\n",
2828
+ " 50257\n",
2829
+ " ],\n",
2830
+ " \"bos_token_id\": 50257,\n",
2831
+ " \"decoder_start_token_id\": 50258,\n",
2832
+ " \"eos_token_id\": 50257,\n",
2833
+ " \"max_length\": 448,\n",
2834
+ " \"pad_token_id\": 50257,\n",
2835
+ " \"suppress_tokens\": [],\n",
2836
+ " \"transformers_version\": \"4.26.0.dev0\",\n",
2837
+ " \"use_cache\": false\n",
2838
+ "}\n",
2839
+ "\n",
2840
+ "Generate config GenerationConfig {\n",
2841
+ " \"begin_suppress_tokens\": [\n",
2842
+ " 220,\n",
2843
+ " 50257\n",
2844
+ " ],\n",
2845
+ " \"bos_token_id\": 50257,\n",
2846
+ " \"decoder_start_token_id\": 50258,\n",
2847
+ " \"eos_token_id\": 50257,\n",
2848
+ " \"max_length\": 448,\n",
2849
+ " \"pad_token_id\": 50257,\n",
2850
+ " \"suppress_tokens\": [],\n",
2851
+ " \"transformers_version\": \"4.26.0.dev0\",\n",
2852
+ " \"use_cache\": false\n",
2853
+ "}\n",
2854
+ "\n",
2855
+ "Generate config GenerationConfig {\n",
2856
+ " \"begin_suppress_tokens\": [\n",
2857
+ " 220,\n",
2858
+ " 50257\n",
2859
+ " ],\n",
2860
+ " \"bos_token_id\": 50257,\n",
2861
+ " \"decoder_start_token_id\": 50258,\n",
2862
+ " \"eos_token_id\": 50257,\n",
2863
+ " \"max_length\": 448,\n",
2864
+ " \"pad_token_id\": 50257,\n",
2865
+ " \"suppress_tokens\": [],\n",
2866
+ " \"transformers_version\": \"4.26.0.dev0\",\n",
2867
+ " \"use_cache\": false\n",
2868
+ "}\n",
2869
+ "\n",
2870
+ "Generate config GenerationConfig {\n",
2871
+ " \"begin_suppress_tokens\": [\n",
2872
+ " 220,\n",
2873
+ " 50257\n",
2874
+ " ],\n",
2875
+ " \"bos_token_id\": 50257,\n",
2876
+ " \"decoder_start_token_id\": 50258,\n",
2877
+ " \"eos_token_id\": 50257,\n",
2878
+ " \"max_length\": 448,\n",
2879
+ " \"pad_token_id\": 50257,\n",
2880
+ " \"suppress_tokens\": [],\n",
2881
+ " \"transformers_version\": \"4.26.0.dev0\",\n",
2882
+ " \"use_cache\": false\n",
2883
+ "}\n",
2884
+ "\n",
2885
+ "Generate config GenerationConfig {\n",
2886
+ " \"begin_suppress_tokens\": [\n",
2887
+ " 220,\n",
2888
+ " 50257\n",
2889
+ " ],\n",
2890
+ " \"bos_token_id\": 50257,\n",
2891
+ " \"decoder_start_token_id\": 50258,\n",
2892
+ " \"eos_token_id\": 50257,\n",
2893
+ " \"max_length\": 448,\n",
2894
+ " \"pad_token_id\": 50257,\n",
2895
+ " \"suppress_tokens\": [],\n",
2896
+ " \"transformers_version\": \"4.26.0.dev0\",\n",
2897
+ " \"use_cache\": false\n",
2898
+ "}\n",
2899
+ "\n",
2900
+ "Generate config GenerationConfig {\n",
2901
+ " \"begin_suppress_tokens\": [\n",
2902
+ " 220,\n",
2903
+ " 50257\n",
2904
+ " ],\n",
2905
+ " \"bos_token_id\": 50257,\n",
2906
+ " \"decoder_start_token_id\": 50258,\n",
2907
+ " \"eos_token_id\": 50257,\n",
2908
+ " \"max_length\": 448,\n",
2909
+ " \"pad_token_id\": 50257,\n",
2910
+ " \"suppress_tokens\": [],\n",
2911
+ " \"transformers_version\": \"4.26.0.dev0\",\n",
2912
+ " \"use_cache\": false\n",
2913
+ "}\n",
2914
+ "\n",
2915
+ "Generate config GenerationConfig {\n",
2916
+ " \"begin_suppress_tokens\": [\n",
2917
+ " 220,\n",
2918
+ " 50257\n",
2919
+ " ],\n",
2920
+ " \"bos_token_id\": 50257,\n",
2921
+ " \"decoder_start_token_id\": 50258,\n",
2922
+ " \"eos_token_id\": 50257,\n",
2923
+ " \"max_length\": 448,\n",
2924
+ " \"pad_token_id\": 50257,\n",
2925
+ " \"suppress_tokens\": [],\n",
2926
+ " \"transformers_version\": \"4.26.0.dev0\",\n",
2927
+ " \"use_cache\": false\n",
2928
+ "}\n",
2929
+ "\n",
2930
+ "Generate config GenerationConfig {\n",
2931
+ " \"begin_suppress_tokens\": [\n",
2932
+ " 220,\n",
2933
+ " 50257\n",
2934
+ " ],\n",
2935
+ " \"bos_token_id\": 50257,\n",
2936
+ " \"decoder_start_token_id\": 50258,\n",
2937
+ " \"eos_token_id\": 50257,\n",
2938
+ " \"max_length\": 448,\n",
2939
+ " \"pad_token_id\": 50257,\n",
2940
+ " \"suppress_tokens\": [],\n",
2941
+ " \"transformers_version\": \"4.26.0.dev0\",\n",
2942
+ " \"use_cache\": false\n",
2943
+ "}\n",
2944
+ "\n",
2945
+ "Generate config GenerationConfig {\n",
2946
+ " \"begin_suppress_tokens\": [\n",
2947
+ " 220,\n",
2948
+ " 50257\n",
2949
+ " ],\n",
2950
+ " \"bos_token_id\": 50257,\n",
2951
+ " \"decoder_start_token_id\": 50258,\n",
2952
+ " \"eos_token_id\": 50257,\n",
2953
+ " \"max_length\": 448,\n",
2954
+ " \"pad_token_id\": 50257,\n",
2955
+ " \"suppress_tokens\": [],\n",
2956
+ " \"transformers_version\": \"4.26.0.dev0\",\n",
2957
+ " \"use_cache\": false\n",
2958
+ "}\n",
2959
+ "\n",
2960
+ "Generate config GenerationConfig {\n",
2961
+ " \"begin_suppress_tokens\": [\n",
2962
+ " 220,\n",
2963
+ " 50257\n",
2964
+ " ],\n",
2965
+ " \"bos_token_id\": 50257,\n",
2966
+ " \"decoder_start_token_id\": 50258,\n",
2967
+ " \"eos_token_id\": 50257,\n",
2968
+ " \"max_length\": 448,\n",
2969
+ " \"pad_token_id\": 50257,\n",
2970
+ " \"suppress_tokens\": [],\n",
2971
+ " \"transformers_version\": \"4.26.0.dev0\",\n",
2972
+ " \"use_cache\": false\n",
2973
+ "}\n",
2974
+ "\n",
2975
+ "Generate config GenerationConfig {\n",
2976
+ " \"begin_suppress_tokens\": [\n",
2977
+ " 220,\n",
2978
+ " 50257\n",
2979
+ " ],\n",
2980
+ " \"bos_token_id\": 50257,\n",
2981
+ " \"decoder_start_token_id\": 50258,\n",
2982
+ " \"eos_token_id\": 50257,\n",
2983
+ " \"max_length\": 448,\n",
2984
+ " \"pad_token_id\": 50257,\n",
2985
+ " \"suppress_tokens\": [],\n",
2986
+ " \"transformers_version\": \"4.26.0.dev0\",\n",
2987
+ " \"use_cache\": false\n",
2988
+ "}\n",
2989
+ "\n",
2990
+ "Generate config GenerationConfig {\n",
2991
+ " \"begin_suppress_tokens\": [\n",
2992
+ " 220,\n",
2993
+ " 50257\n",
2994
+ " ],\n",
2995
+ " \"bos_token_id\": 50257,\n",
2996
+ " \"decoder_start_token_id\": 50258,\n",
2997
+ " \"eos_token_id\": 50257,\n",
2998
+ " \"max_length\": 448,\n",
2999
+ " \"pad_token_id\": 50257,\n",
3000
+ " \"suppress_tokens\": [],\n",
3001
+ " \"transformers_version\": \"4.26.0.dev0\",\n",
3002
+ " \"use_cache\": false\n",
3003
+ "}\n",
3004
+ "\n",
3005
+ "Generate config GenerationConfig {\n",
3006
+ " \"begin_suppress_tokens\": [\n",
3007
+ " 220,\n",
3008
+ " 50257\n",
3009
+ " ],\n",
3010
+ " \"bos_token_id\": 50257,\n",
3011
+ " \"decoder_start_token_id\": 50258,\n",
3012
+ " \"eos_token_id\": 50257,\n",
3013
+ " \"max_length\": 448,\n",
3014
+ " \"pad_token_id\": 50257,\n",
3015
+ " \"suppress_tokens\": [],\n",
3016
+ " \"transformers_version\": \"4.26.0.dev0\",\n",
3017
+ " \"use_cache\": false\n",
3018
+ "}\n",
3019
+ "\n",
3020
+ "Generate config GenerationConfig {\n",
3021
+ " \"begin_suppress_tokens\": [\n",
3022
+ " 220,\n",
3023
+ " 50257\n",
3024
+ " ],\n",
3025
+ " \"bos_token_id\": 50257,\n",
3026
+ " \"decoder_start_token_id\": 50258,\n",
3027
+ " \"eos_token_id\": 50257,\n",
3028
+ " \"max_length\": 448,\n",
3029
+ " \"pad_token_id\": 50257,\n",
3030
+ " \"suppress_tokens\": [],\n",
3031
+ " \"transformers_version\": \"4.26.0.dev0\",\n",
3032
+ " \"use_cache\": false\n",
3033
+ "}\n",
3034
+ "\n",
3035
+ "Generate config GenerationConfig {\n",
3036
+ " \"begin_suppress_tokens\": [\n",
3037
+ " 220,\n",
3038
+ " 50257\n",
3039
+ " ],\n",
3040
+ " \"bos_token_id\": 50257,\n",
3041
+ " \"decoder_start_token_id\": 50258,\n",
3042
+ " \"eos_token_id\": 50257,\n",
3043
+ " \"max_length\": 448,\n",
3044
+ " \"pad_token_id\": 50257,\n",
3045
+ " \"suppress_tokens\": [],\n",
3046
+ " \"transformers_version\": \"4.26.0.dev0\",\n",
3047
+ " \"use_cache\": false\n",
3048
+ "}\n",
3049
+ "\n",
3050
+ "Generate config GenerationConfig {\n",
3051
+ " \"begin_suppress_tokens\": [\n",
3052
+ " 220,\n",
3053
+ " 50257\n",
3054
+ " ],\n",
3055
+ " \"bos_token_id\": 50257,\n",
3056
+ " \"decoder_start_token_id\": 50258,\n",
3057
+ " \"eos_token_id\": 50257,\n",
3058
+ " \"max_length\": 448,\n",
3059
+ " \"pad_token_id\": 50257,\n",
3060
+ " \"suppress_tokens\": [],\n",
3061
+ " \"transformers_version\": \"4.26.0.dev0\",\n",
3062
+ " \"use_cache\": false\n",
3063
+ "}\n",
3064
+ "\n",
3065
+ "Generate config GenerationConfig {\n",
3066
+ " \"begin_suppress_tokens\": [\n",
3067
+ " 220,\n",
3068
+ " 50257\n",
3069
+ " ],\n",
3070
+ " \"bos_token_id\": 50257,\n",
3071
+ " \"decoder_start_token_id\": 50258,\n",
3072
+ " \"eos_token_id\": 50257,\n",
3073
+ " \"max_length\": 448,\n",
3074
+ " \"pad_token_id\": 50257,\n",
3075
+ " \"suppress_tokens\": [],\n",
3076
+ " \"transformers_version\": \"4.26.0.dev0\",\n",
3077
+ " \"use_cache\": false\n",
3078
+ "}\n",
3079
+ "\n",
3080
+ "Generate config GenerationConfig {\n",
3081
+ " \"begin_suppress_tokens\": [\n",
3082
+ " 220,\n",
3083
+ " 50257\n",
3084
+ " ],\n",
3085
+ " \"bos_token_id\": 50257,\n",
3086
+ " \"decoder_start_token_id\": 50258,\n",
3087
+ " \"eos_token_id\": 50257,\n",
3088
+ " \"max_length\": 448,\n",
3089
+ " \"pad_token_id\": 50257,\n",
3090
+ " \"suppress_tokens\": [],\n",
3091
+ " \"transformers_version\": \"4.26.0.dev0\",\n",
3092
+ " \"use_cache\": false\n",
3093
+ "}\n",
3094
+ "\n",
3095
+ "Generate config GenerationConfig {\n",
3096
+ " \"begin_suppress_tokens\": [\n",
3097
+ " 220,\n",
3098
+ " 50257\n",
3099
+ " ],\n",
3100
+ " \"bos_token_id\": 50257,\n",
3101
+ " \"decoder_start_token_id\": 50258,\n",
3102
+ " \"eos_token_id\": 50257,\n",
3103
+ " \"max_length\": 448,\n",
3104
+ " \"pad_token_id\": 50257,\n",
3105
+ " \"suppress_tokens\": [],\n",
3106
+ " \"transformers_version\": \"4.26.0.dev0\",\n",
3107
+ " \"use_cache\": false\n",
3108
+ "}\n",
3109
+ "\n",
3110
+ "Generate config GenerationConfig {\n",
3111
+ " \"begin_suppress_tokens\": [\n",
3112
+ " 220,\n",
3113
+ " 50257\n",
3114
+ " ],\n",
3115
+ " \"bos_token_id\": 50257,\n",
3116
+ " \"decoder_start_token_id\": 50258,\n",
3117
+ " \"eos_token_id\": 50257,\n",
3118
+ " \"max_length\": 448,\n",
3119
+ " \"pad_token_id\": 50257,\n",
3120
+ " \"suppress_tokens\": [],\n",
3121
+ " \"transformers_version\": \"4.26.0.dev0\",\n",
3122
+ " \"use_cache\": false\n",
3123
+ "}\n",
3124
+ "\n",
3125
+ "Generate config GenerationConfig {\n",
3126
+ " \"begin_suppress_tokens\": [\n",
3127
+ " 220,\n",
3128
+ " 50257\n",
3129
+ " ],\n",
3130
+ " \"bos_token_id\": 50257,\n",
3131
+ " \"decoder_start_token_id\": 50258,\n",
3132
+ " \"eos_token_id\": 50257,\n",
3133
+ " \"max_length\": 448,\n",
3134
+ " \"pad_token_id\": 50257,\n",
3135
+ " \"suppress_tokens\": [],\n",
3136
+ " \"transformers_version\": \"4.26.0.dev0\",\n",
3137
+ " \"use_cache\": false\n",
3138
+ "}\n",
3139
+ "\n",
3140
+ "Generate config GenerationConfig {\n",
3141
+ " \"begin_suppress_tokens\": [\n",
3142
+ " 220,\n",
3143
+ " 50257\n",
3144
+ " ],\n",
3145
+ " \"bos_token_id\": 50257,\n",
3146
+ " \"decoder_start_token_id\": 50258,\n",
3147
+ " \"eos_token_id\": 50257,\n",
3148
+ " \"max_length\": 448,\n",
3149
+ " \"pad_token_id\": 50257,\n",
3150
+ " \"suppress_tokens\": [],\n",
3151
+ " \"transformers_version\": \"4.26.0.dev0\",\n",
3152
+ " \"use_cache\": false\n",
3153
+ "}\n",
3154
+ "\n"
3155
+ ]
3156
+ },
3157
+ {
3158
+ "name": "stderr",
3159
+ "output_type": "stream",
3160
+ "text": [
3161
+ "Generate config GenerationConfig {\n",
3162
+ " \"begin_suppress_tokens\": [\n",
3163
+ " 220,\n",
3164
+ " 50257\n",
3165
+ " ],\n",
3166
+ " \"bos_token_id\": 50257,\n",
3167
+ " \"decoder_start_token_id\": 50258,\n",
3168
+ " \"eos_token_id\": 50257,\n",
3169
+ " \"max_length\": 448,\n",
3170
+ " \"pad_token_id\": 50257,\n",
3171
+ " \"suppress_tokens\": [],\n",
3172
+ " \"transformers_version\": \"4.26.0.dev0\",\n",
3173
+ " \"use_cache\": false\n",
3174
+ "}\n",
3175
+ "\n",
3176
+ "Generate config GenerationConfig {\n",
3177
+ " \"begin_suppress_tokens\": [\n",
3178
+ " 220,\n",
3179
+ " 50257\n",
3180
+ " ],\n",
3181
+ " \"bos_token_id\": 50257,\n",
3182
+ " \"decoder_start_token_id\": 50258,\n",
3183
+ " \"eos_token_id\": 50257,\n",
3184
+ " \"max_length\": 448,\n",
3185
+ " \"pad_token_id\": 50257,\n",
3186
+ " \"suppress_tokens\": [],\n",
3187
+ " \"transformers_version\": \"4.26.0.dev0\",\n",
3188
+ " \"use_cache\": false\n",
3189
+ "}\n",
3190
+ "\n",
3191
+ "Generate config GenerationConfig {\n",
3192
+ " \"begin_suppress_tokens\": [\n",
3193
+ " 220,\n",
3194
+ " 50257\n",
3195
+ " ],\n",
3196
+ " \"bos_token_id\": 50257,\n",
3197
+ " \"decoder_start_token_id\": 50258,\n",
3198
+ " \"eos_token_id\": 50257,\n",
3199
+ " \"max_length\": 448,\n",
3200
+ " \"pad_token_id\": 50257,\n",
3201
+ " \"suppress_tokens\": [],\n",
3202
+ " \"transformers_version\": \"4.26.0.dev0\",\n",
3203
+ " \"use_cache\": false\n",
3204
+ "}\n",
3205
+ "\n",
3206
+ "Generate config GenerationConfig {\n",
3207
+ " \"begin_suppress_tokens\": [\n",
3208
+ " 220,\n",
3209
+ " 50257\n",
3210
+ " ],\n",
3211
+ " \"bos_token_id\": 50257,\n",
3212
+ " \"decoder_start_token_id\": 50258,\n",
3213
+ " \"eos_token_id\": 50257,\n",
3214
+ " \"max_length\": 448,\n",
3215
+ " \"pad_token_id\": 50257,\n",
3216
+ " \"suppress_tokens\": [],\n",
3217
+ " \"transformers_version\": \"4.26.0.dev0\",\n",
3218
+ " \"use_cache\": false\n",
3219
+ "}\n",
3220
+ "\n",
3221
+ "Generate config GenerationConfig {\n",
3222
+ " \"begin_suppress_tokens\": [\n",
3223
+ " 220,\n",
3224
+ " 50257\n",
3225
+ " ],\n",
3226
+ " \"bos_token_id\": 50257,\n",
3227
+ " \"decoder_start_token_id\": 50258,\n",
3228
+ " \"eos_token_id\": 50257,\n",
3229
+ " \"max_length\": 448,\n",
3230
+ " \"pad_token_id\": 50257,\n",
3231
+ " \"suppress_tokens\": [],\n",
3232
+ " \"transformers_version\": \"4.26.0.dev0\",\n",
3233
+ " \"use_cache\": false\n",
3234
+ "}\n",
3235
+ "\n",
3236
+ "Generate config GenerationConfig {\n",
3237
+ " \"begin_suppress_tokens\": [\n",
3238
+ " 220,\n",
3239
+ " 50257\n",
3240
+ " ],\n",
3241
+ " \"bos_token_id\": 50257,\n",
3242
+ " \"decoder_start_token_id\": 50258,\n",
3243
+ " \"eos_token_id\": 50257,\n",
3244
+ " \"max_length\": 448,\n",
3245
+ " \"pad_token_id\": 50257,\n",
3246
+ " \"suppress_tokens\": [],\n",
3247
+ " \"transformers_version\": \"4.26.0.dev0\",\n",
3248
+ " \"use_cache\": false\n",
3249
+ "}\n",
3250
+ "\n",
3251
+ "Generate config GenerationConfig {\n",
3252
+ " \"begin_suppress_tokens\": [\n",
3253
+ " 220,\n",
3254
+ " 50257\n",
3255
+ " ],\n",
3256
+ " \"bos_token_id\": 50257,\n",
3257
+ " \"decoder_start_token_id\": 50258,\n",
3258
+ " \"eos_token_id\": 50257,\n",
3259
+ " \"max_length\": 448,\n",
3260
+ " \"pad_token_id\": 50257,\n",
3261
+ " \"suppress_tokens\": [],\n",
3262
+ " \"transformers_version\": \"4.26.0.dev0\",\n",
3263
+ " \"use_cache\": false\n",
3264
+ "}\n",
3265
+ "\n",
3266
+ "Generate config GenerationConfig {\n",
3267
+ " \"begin_suppress_tokens\": [\n",
3268
+ " 220,\n",
3269
+ " 50257\n",
3270
+ " ],\n",
3271
+ " \"bos_token_id\": 50257,\n",
3272
+ " \"decoder_start_token_id\": 50258,\n",
3273
+ " \"eos_token_id\": 50257,\n",
3274
+ " \"max_length\": 448,\n",
3275
+ " \"pad_token_id\": 50257,\n",
3276
+ " \"suppress_tokens\": [],\n",
3277
+ " \"transformers_version\": \"4.26.0.dev0\",\n",
3278
+ " \"use_cache\": false\n",
3279
+ "}\n",
3280
+ "\n",
3281
+ "Generate config GenerationConfig {\n",
3282
+ " \"begin_suppress_tokens\": [\n",
3283
+ " 220,\n",
3284
+ " 50257\n",
3285
+ " ],\n",
3286
+ " \"bos_token_id\": 50257,\n",
3287
+ " \"decoder_start_token_id\": 50258,\n",
3288
+ " \"eos_token_id\": 50257,\n",
3289
+ " \"max_length\": 448,\n",
3290
+ " \"pad_token_id\": 50257,\n",
3291
+ " \"suppress_tokens\": [],\n",
3292
+ " \"transformers_version\": \"4.26.0.dev0\",\n",
3293
+ " \"use_cache\": false\n",
3294
+ "}\n",
3295
+ "\n",
3296
+ "Generate config GenerationConfig {\n",
3297
+ " \"begin_suppress_tokens\": [\n",
3298
+ " 220,\n",
3299
+ " 50257\n",
3300
+ " ],\n",
3301
+ " \"bos_token_id\": 50257,\n",
3302
+ " \"decoder_start_token_id\": 50258,\n",
3303
+ " \"eos_token_id\": 50257,\n",
3304
+ " \"max_length\": 448,\n",
3305
+ " \"pad_token_id\": 50257,\n",
3306
+ " \"suppress_tokens\": [],\n",
3307
+ " \"transformers_version\": \"4.26.0.dev0\",\n",
3308
+ " \"use_cache\": false\n",
3309
+ "}\n",
3310
+ "\n",
3311
+ "Generate config GenerationConfig {\n",
3312
+ " \"begin_suppress_tokens\": [\n",
3313
+ " 220,\n",
3314
+ " 50257\n",
3315
+ " ],\n",
3316
+ " \"bos_token_id\": 50257,\n",
3317
+ " \"decoder_start_token_id\": 50258,\n",
3318
+ " \"eos_token_id\": 50257,\n",
3319
+ " \"max_length\": 448,\n",
3320
+ " \"pad_token_id\": 50257,\n",
3321
+ " \"suppress_tokens\": [],\n",
3322
+ " \"transformers_version\": \"4.26.0.dev0\",\n",
3323
+ " \"use_cache\": false\n",
3324
+ "}\n",
3325
+ "\n",
3326
+ "Generate config GenerationConfig {\n",
3327
+ " \"begin_suppress_tokens\": [\n",
3328
+ " 220,\n",
3329
+ " 50257\n",
3330
+ " ],\n",
3331
+ " \"bos_token_id\": 50257,\n",
3332
+ " \"decoder_start_token_id\": 50258,\n",
3333
+ " \"eos_token_id\": 50257,\n",
3334
+ " \"max_length\": 448,\n",
3335
+ " \"pad_token_id\": 50257,\n",
3336
+ " \"suppress_tokens\": [],\n",
3337
+ " \"transformers_version\": \"4.26.0.dev0\",\n",
3338
+ " \"use_cache\": false\n",
3339
+ "}\n",
3340
+ "\n",
3341
+ "Generate config GenerationConfig {\n",
3342
+ " \"begin_suppress_tokens\": [\n",
3343
+ " 220,\n",
3344
+ " 50257\n",
3345
+ " ],\n",
3346
+ " \"bos_token_id\": 50257,\n",
3347
+ " \"decoder_start_token_id\": 50258,\n",
3348
+ " \"eos_token_id\": 50257,\n",
3349
+ " \"max_length\": 448,\n",
3350
+ " \"pad_token_id\": 50257,\n",
3351
+ " \"suppress_tokens\": [],\n",
3352
+ " \"transformers_version\": \"4.26.0.dev0\",\n",
3353
+ " \"use_cache\": false\n",
3354
+ "}\n",
3355
+ "\n",
3356
+ "Generate config GenerationConfig {\n",
3357
+ " \"begin_suppress_tokens\": [\n",
3358
+ " 220,\n",
3359
+ " 50257\n",
3360
+ " ],\n",
3361
+ " \"bos_token_id\": 50257,\n",
3362
+ " \"decoder_start_token_id\": 50258,\n",
3363
+ " \"eos_token_id\": 50257,\n",
3364
+ " \"max_length\": 448,\n",
3365
+ " \"pad_token_id\": 50257,\n",
3366
+ " \"suppress_tokens\": [],\n",
3367
+ " \"transformers_version\": \"4.26.0.dev0\",\n",
3368
+ " \"use_cache\": false\n",
3369
+ "}\n",
3370
+ "\n",
3371
+ "Generate config GenerationConfig {\n",
3372
+ " \"begin_suppress_tokens\": [\n",
3373
+ " 220,\n",
3374
+ " 50257\n",
3375
+ " ],\n",
3376
+ " \"bos_token_id\": 50257,\n",
3377
+ " \"decoder_start_token_id\": 50258,\n",
3378
+ " \"eos_token_id\": 50257,\n",
3379
+ " \"max_length\": 448,\n",
3380
+ " \"pad_token_id\": 50257,\n",
3381
+ " \"suppress_tokens\": [],\n",
3382
+ " \"transformers_version\": \"4.26.0.dev0\",\n",
3383
+ " \"use_cache\": false\n",
3384
+ "}\n",
3385
+ "\n",
3386
+ "Generate config GenerationConfig {\n",
3387
+ " \"begin_suppress_tokens\": [\n",
3388
+ " 220,\n",
3389
+ " 50257\n",
3390
+ " ],\n",
3391
+ " \"bos_token_id\": 50257,\n",
3392
+ " \"decoder_start_token_id\": 50258,\n",
3393
+ " \"eos_token_id\": 50257,\n",
3394
+ " \"max_length\": 448,\n",
3395
+ " \"pad_token_id\": 50257,\n",
3396
+ " \"suppress_tokens\": [],\n",
3397
+ " \"transformers_version\": \"4.26.0.dev0\",\n",
3398
+ " \"use_cache\": false\n",
3399
+ "}\n",
3400
+ "\n",
3401
+ "Generate config GenerationConfig {\n",
3402
+ " \"begin_suppress_tokens\": [\n",
3403
+ " 220,\n",
3404
+ " 50257\n",
3405
+ " ],\n",
3406
+ " \"bos_token_id\": 50257,\n",
3407
+ " \"decoder_start_token_id\": 50258,\n",
3408
+ " \"eos_token_id\": 50257,\n",
3409
+ " \"max_length\": 448,\n",
3410
+ " \"pad_token_id\": 50257,\n",
3411
+ " \"suppress_tokens\": [],\n",
3412
+ " \"transformers_version\": \"4.26.0.dev0\",\n",
3413
+ " \"use_cache\": false\n",
3414
+ "}\n",
3415
+ "\n",
3416
+ "Generate config GenerationConfig {\n",
3417
+ " \"begin_suppress_tokens\": [\n",
3418
+ " 220,\n",
3419
+ " 50257\n",
3420
+ " ],\n",
3421
+ " \"bos_token_id\": 50257,\n",
3422
+ " \"decoder_start_token_id\": 50258,\n",
3423
+ " \"eos_token_id\": 50257,\n",
3424
+ " \"max_length\": 448,\n",
3425
+ " \"pad_token_id\": 50257,\n",
3426
+ " \"suppress_tokens\": [],\n",
3427
+ " \"transformers_version\": \"4.26.0.dev0\",\n",
3428
+ " \"use_cache\": false\n",
3429
+ "}\n",
3430
+ "\n",
3431
+ "Generate config GenerationConfig {\n",
3432
+ " \"begin_suppress_tokens\": [\n",
3433
+ " 220,\n",
3434
+ " 50257\n",
3435
+ " ],\n",
3436
+ " \"bos_token_id\": 50257,\n",
3437
+ " \"decoder_start_token_id\": 50258,\n",
3438
+ " \"eos_token_id\": 50257,\n",
3439
+ " \"max_length\": 448,\n",
3440
+ " \"pad_token_id\": 50257,\n",
3441
+ " \"suppress_tokens\": [],\n",
3442
+ " \"transformers_version\": \"4.26.0.dev0\",\n",
3443
+ " \"use_cache\": false\n",
3444
+ "}\n",
3445
+ "\n",
3446
+ "Generate config GenerationConfig {\n",
3447
+ " \"begin_suppress_tokens\": [\n",
3448
+ " 220,\n",
3449
+ " 50257\n",
3450
+ " ],\n",
3451
+ " \"bos_token_id\": 50257,\n",
3452
+ " \"decoder_start_token_id\": 50258,\n",
3453
+ " \"eos_token_id\": 50257,\n",
3454
+ " \"max_length\": 448,\n",
3455
+ " \"pad_token_id\": 50257,\n",
3456
+ " \"suppress_tokens\": [],\n",
3457
+ " \"transformers_version\": \"4.26.0.dev0\",\n",
3458
+ " \"use_cache\": false\n",
3459
+ "}\n",
3460
+ "\n",
3461
+ "Generate config GenerationConfig {\n",
3462
+ " \"begin_suppress_tokens\": [\n",
3463
+ " 220,\n",
3464
+ " 50257\n",
3465
+ " ],\n",
3466
+ " \"bos_token_id\": 50257,\n",
3467
+ " \"decoder_start_token_id\": 50258,\n",
3468
+ " \"eos_token_id\": 50257,\n",
3469
+ " \"max_length\": 448,\n",
3470
+ " \"pad_token_id\": 50257,\n",
3471
+ " \"suppress_tokens\": [],\n",
3472
+ " \"transformers_version\": \"4.26.0.dev0\",\n",
3473
+ " \"use_cache\": false\n",
3474
+ "}\n",
3475
+ "\n",
3476
+ "Generate config GenerationConfig {\n",
3477
+ " \"begin_suppress_tokens\": [\n",
3478
+ " 220,\n",
3479
+ " 50257\n",
3480
+ " ],\n",
3481
+ " \"bos_token_id\": 50257,\n",
3482
+ " \"decoder_start_token_id\": 50258,\n",
3483
+ " \"eos_token_id\": 50257,\n",
3484
+ " \"max_length\": 448,\n",
3485
+ " \"pad_token_id\": 50257,\n",
3486
+ " \"suppress_tokens\": [],\n",
3487
+ " \"transformers_version\": \"4.26.0.dev0\",\n",
3488
+ " \"use_cache\": false\n",
3489
+ "}\n",
3490
+ "\n",
3491
+ "Saving model checkpoint to ./checkpoint-1000\n",
3492
+ "Configuration saved in ./checkpoint-1000/config.json\n",
3493
+ "Model weights saved in ./checkpoint-1000/pytorch_model.bin\n",
3494
+ "Feature extractor saved in ./checkpoint-1000/preprocessor_config.json\n",
3495
+ "Feature extractor saved in ./preprocessor_config.json\n"
3496
+ ]
3497
  }
3498
  ],
3499
  "source": [
fine-tune-whisper-non-streaming-fleurs-ms.ipynb CHANGED
@@ -2738,8 +2738,8 @@
2738
  "\n",
2739
  " <div>\n",
2740
  " \n",
2741
- " <progress value='1001' max='5000' style='width:300px; height:20px; vertical-align: middle;'></progress>\n",
2742
- " [1001/5000 1:51:25 < 7:26:02, 0.15 it/s, Epoch 10.64/54]\n",
2743
  " </div>\n",
2744
  " <table border=\"1\" class=\"dataframe\">\n",
2745
  " <thead>\n",
@@ -2747,17 +2747,18 @@
2747
  " <th>Step</th>\n",
2748
  " <th>Training Loss</th>\n",
2749
  " <th>Validation Loss</th>\n",
 
2750
  " </tr>\n",
2751
  " </thead>\n",
2752
  " <tbody>\n",
 
 
 
 
 
 
2753
  " </tbody>\n",
2754
- "</table><p>\n",
2755
- " <div>\n",
2756
- " \n",
2757
- " <progress value='47' max='47' style='width:300px; height:20px; vertical-align: middle;'></progress>\n",
2758
- " [47/47 15:11]\n",
2759
- " </div>\n",
2760
- " "
2761
  ],
2762
  "text/plain": [
2763
  "<IPython.core.display.HTML object>"
@@ -3486,7 +3487,12 @@
3486
  " \"transformers_version\": \"4.26.0.dev0\",\n",
3487
  " \"use_cache\": false\n",
3488
  "}\n",
3489
- "\n"
 
 
 
 
 
3490
  ]
3491
  }
3492
  ],
 
2738
  "\n",
2739
  " <div>\n",
2740
  " \n",
2741
+ " <progress value='1028' max='5000' style='width:300px; height:20px; vertical-align: middle;'></progress>\n",
2742
+ " [1028/5000 2:10:54 < 8:26:47, 0.13 it/s, Epoch 10.93/54]\n",
2743
  " </div>\n",
2744
  " <table border=\"1\" class=\"dataframe\">\n",
2745
  " <thead>\n",
 
2747
  " <th>Step</th>\n",
2748
  " <th>Training Loss</th>\n",
2749
  " <th>Validation Loss</th>\n",
2750
+ " <th>Wer</th>\n",
2751
  " </tr>\n",
2752
  " </thead>\n",
2753
  " <tbody>\n",
2754
+ " <tr>\n",
2755
+ " <td>1000</td>\n",
2756
+ " <td>0.002400</td>\n",
2757
+ " <td>0.243783</td>\n",
2758
+ " <td>10.344360</td>\n",
2759
+ " </tr>\n",
2760
  " </tbody>\n",
2761
+ "</table><p>"
 
 
 
 
 
 
2762
  ],
2763
  "text/plain": [
2764
  "<IPython.core.display.HTML object>"
 
3487
  " \"transformers_version\": \"4.26.0.dev0\",\n",
3488
  " \"use_cache\": false\n",
3489
  "}\n",
3490
+ "\n",
3491
+ "Saving model checkpoint to ./checkpoint-1000\n",
3492
+ "Configuration saved in ./checkpoint-1000/config.json\n",
3493
+ "Model weights saved in ./checkpoint-1000/pytorch_model.bin\n",
3494
+ "Feature extractor saved in ./checkpoint-1000/preprocessor_config.json\n",
3495
+ "Feature extractor saved in ./preprocessor_config.json\n"
3496
  ]
3497
  }
3498
  ],
pytorch_model.bin CHANGED
@@ -1,3 +1,3 @@
1
  version https://git-lfs.github.com/spec/v1
2
- oid sha256:daf516ed845916ed13cb742b54efbf9a92c6d617418efdca4e9f9fbe273ddc0b
3
  size 3055754841
 
1
  version https://git-lfs.github.com/spec/v1
2
+ oid sha256:e48224641d4c07badfad88bf64aaa85944bcc73d30ed21853cb493513c2d380d
3
  size 3055754841
runs/Dec16_12-04-12_129-213-26-202/events.out.tfevents.1671192266.129-213-26-202.127891.0 CHANGED
@@ -1,3 +1,3 @@
1
  version https://git-lfs.github.com/spec/v1
2
- oid sha256:76cc2c6a7dba75a994293da0acd1f1182b0285542cccd0f1595a46cead8110c1
3
- size 10873
 
1
  version https://git-lfs.github.com/spec/v1
2
+ oid sha256:73d51e9901db41f34b5c143ae7ab2efb2a551f7f6142b3fd4161dd7d7cb079a9
3
+ size 17471