Publish current best Formosan ASR checkpoint for Saisiyat

.gitattributes

@@ -1,35 +1 @@
-*.7z filter=lfs diff=lfs merge=lfs -text
-*.arrow filter=lfs diff=lfs merge=lfs -text
-*.bin filter=lfs diff=lfs merge=lfs -text
-*.bz2 filter=lfs diff=lfs merge=lfs -text
-*.ckpt filter=lfs diff=lfs merge=lfs -text
-*.ftz filter=lfs diff=lfs merge=lfs -text
-*.gz filter=lfs diff=lfs merge=lfs -text
-*.h5 filter=lfs diff=lfs merge=lfs -text
-*.joblib filter=lfs diff=lfs merge=lfs -text
-*.lfs.* filter=lfs diff=lfs merge=lfs -text
-*.mlmodel filter=lfs diff=lfs merge=lfs -text
-*.model filter=lfs diff=lfs merge=lfs -text
-*.msgpack filter=lfs diff=lfs merge=lfs -text
-*.npy filter=lfs diff=lfs merge=lfs -text
-*.npz filter=lfs diff=lfs merge=lfs -text
-*.onnx filter=lfs diff=lfs merge=lfs -text
-*.ot filter=lfs diff=lfs merge=lfs -text
-*.parquet filter=lfs diff=lfs merge=lfs -text
-*.pb filter=lfs diff=lfs merge=lfs -text
-*.pickle filter=lfs diff=lfs merge=lfs -text
-*.pkl filter=lfs diff=lfs merge=lfs -text
-*.pt filter=lfs diff=lfs merge=lfs -text
-*.pth filter=lfs diff=lfs merge=lfs -text
-*.rar filter=lfs diff=lfs merge=lfs -text
 *.safetensors filter=lfs diff=lfs merge=lfs -text
-saved_model/**/* filter=lfs diff=lfs merge=lfs -text
-*.tar.* filter=lfs diff=lfs merge=lfs -text
-*.tar filter=lfs diff=lfs merge=lfs -text
-*.tflite filter=lfs diff=lfs merge=lfs -text
-*.tgz filter=lfs diff=lfs merge=lfs -text
-*.wasm filter=lfs diff=lfs merge=lfs -text
-*.xz filter=lfs diff=lfs merge=lfs -text
-*.zip filter=lfs diff=lfs merge=lfs -text
-*.zst filter=lfs diff=lfs merge=lfs -text
-*tfevents* filter=lfs diff=lfs merge=lfs -text

README.md

@@ -0,0 +1,129 @@
+---
+language:
+- "xsy"
+library_name: transformers
+pipeline_tag: automatic-speech-recognition
+tags:
+- formosanbank
+- formosan
+- endangered-languages
+- speech
+- asr
+- xlsr53
+- tacl
+base_model: facebook/wav2vec2-large-xlsr-53
+base_model_relation: finetune
+license: cc-by-4.0
+metrics:
+- wer
+- cer
+---
+
+# Formosan ASR — Saisiyat
+
+This is FormosanBank's current general-purpose automatic speech recognition
+checkpoint for **Saisiyat**. It was selected by the lowest normalized
+micro word error rate among the completed language-level TACL systems evaluated
+on the same frozen all-corpora test partition.
+
+## Model
+
+- Repository: `FormosanBank/formosan-asr-saisiyat`
+- Architecture family: **XLS-R 53**
+- Training path: **language-pooled supervised fine-tuning**
+- Base model: `facebook/wav2vec2-large-xlsr-53`
+- Parameters are stored in `model.safetensors`
+
+### Training lineage
+
+1. Base checkpoint: `facebook/wav2vec2-large-xlsr-53`.
+2. Supervised ASR fine-tuning on the target language's pooled training corpora.
+
+The target-language supervised stage used `ILRDF`, `NTU`, `ePark1`, `ePark2`, `ePark3`. The experiment used
+`dataset_v1`, frozen `split_v2`, normalization `formosan_safe_v1`, and seed
+`13`. Full machine-readable provenance is included in
+`training_config.json`.
+
+## Evaluation
+
+Evaluation uses the frozen, leakage-controlled `split_v2` test manifests and
+normalized text. Metrics are computed by summing edit counts across all
+available Saisiyat test corpora, so larger corpora contribute in
+proportion to their reference tokens.
+
+- Normalized micro WER: **22.93%**
+- Normalized micro CER: **7.93%**
+- Test utterances: **1,189**
+- Test corpora: **5**
+- Automated readiness: **pass**
+
+| Corpus | Utterances | WER | CER |
+|---|---:|---:|---:|
+| ILRDF | 610 | 21.16% | 8.10% |
+| NTU | 104 | 65.11% | 28.31% |
+| ePark1 | 75 | 15.16% | 2.84% |
+| ePark2 | 106 | 10.68% | 2.62% |
+| ePark3 | 294 | 16.95% | 4.52% |
+
+Detailed edit counts are provided in `evaluation_results.json` and
+`metrics_by_corpus.tsv`.
+
+These figures are specific to the frozen TACL evaluation protocol. They should
+not be compared directly with the April 2026 stage-one releases, which used a
+different experiment snapshot.
+
+## Usage
+
+```python
+from transformers import pipeline
+
+repo_id = "FormosanBank/formosan-asr-saisiyat"
+transcribe = pipeline("automatic-speech-recognition", model=repo_id)
+result = transcribe("path/to/16khz_audio.wav")
+print(result["text"])
+```
+
+For long recordings, segment the audio before inference. The training recipe
+used utterances between 2 and 20 seconds.
+
+## Intended use
+
+This checkpoint supports research, education, language documentation, and
+revitalization work involving Saisiyat. Human review is recommended
+before using transcriptions in archives, publications, teaching materials, or
+other consequential settings.
+
+## Limitations
+
+Performance varies substantially across corpora, speakers, recording
+conditions, dialects, speaking styles, and orthographic conventions. The
+aggregate score can hide weak performance on a particular corpus. The model
+may omit, substitute, or hallucinate words and should not be treated as an
+authoritative transcription source.
+
+## Data provenance
+
+Training and evaluation artifacts come from the frozen Hunter Formosan TACL
+pipeline. Source corpus IDs can include Bible, ILRDF, NTU, ePark1, ePark2,
+ePark3, Xuan, YeddaPalemeqBlog, Youtube, and YutasWilang, depending on language
+availability. Audio is not redistributed in this repository. Underlying
+recordings retain their original rights and access conditions.
+
+## License and attribution
+
+This model release and FormosanBank annotations/metadata are provided under
+**CC BY 4.0**. The upstream `facebook/wav2vec2-large-xlsr-53` checkpoint is distributed under
+Apache 2.0. Users remain responsible for respecting the terms associated with
+underlying source recordings.
+
+Please cite FormosanBank:
+
+```bibtex
+@misc{mohamed2024formosanbank,
+  author = {Mohamed, W. and Le Ferrand, É. and Sung, L.-M. and Prud'hommeaux, E. and Hartshorne, J. K.},
+  title = {FormosanBank},
+  year = {2024},
+  note = {Electronic Resource},
+  url = {https://ai4commsci.gitbook.io/formosanbank}
+}
+```

added_tokens.json

@@ -0,0 +1,4 @@
+{
+  "</s>": 32,
+  "<s>": 31
+}

config.json

@@ -0,0 +1,117 @@
+{
+  "_name_or_path": "facebook/wav2vec2-large-xlsr-53",
+  "activation_dropout": 0.0,
+  "adapter_attn_dim": null,
+  "adapter_kernel_size": 3,
+  "adapter_stride": 2,
+  "add_adapter": false,
+  "apply_spec_augment": true,
+  "architectures": [
+    "Wav2Vec2ForCTC"
+  ],
+  "attention_dropout": 0.0,
+  "bos_token_id": 1,
+  "classifier_proj_size": 256,
+  "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": true,
+  "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": false,
+  "hidden_act": "gelu",
+  "hidden_dropout": 0.0,
+  "hidden_size": 1024,
+  "initializer_range": 0.02,
+  "intermediate_size": 4096,
+  "layer_norm_eps": 1e-05,
+  "layerdrop": 0.0,
+  "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_min_masks": 0,
+  "mask_feature_prob": 0.0,
+  "mask_time_length": 10,
+  "mask_time_min_masks": 2,
+  "mask_time_min_space": 1,
+  "mask_time_other": 0.0,
+  "mask_time_prob": 0.0,
+  "mask_time_selection": "static",
+  "model_type": "wav2vec2",
+  "num_adapter_layers": 3,
+  "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,
+  "output_hidden_size": 1024,
+  "pad_token_id": 30,
+  "proj_codevector_dim": 768,
+  "tdnn_dilation": [
+    1,
+    2,
+    3,
+    1,
+    1
+  ],
+  "tdnn_dim": [
+    512,
+    512,
+    512,
+    512,
+    1500
+  ],
+  "tdnn_kernel": [
+    5,
+    3,
+    3,
+    1,
+    1
+  ],
+  "torch_dtype": "float32",
+  "transformers_version": "4.46.3",
+  "use_weighted_layer_sum": false,
+  "vocab_size": 33,
+  "xvector_output_dim": 512
+}

dev_selection_history.tsv

@@ -0,0 +1,7 @@
+step	train_loss	dev_loss	selection_metric	selection_scope	selection_cer	selection_wer	macro_cell_cer	macro_cell_wer	macro_language_cer	macro_language_wer	micro_cer	micro_wer	dev_cells	dev_languages	dev_rows
+1	21.27985954284668	19.994058513001306	decoded_dev_normalized_cer	macro_cell	0.965583	1.000105	0.965583	1.000105	0.948863	1.000226	0.948863	1.000226	5	1	1190
+7105	0.03559953719377518	0.4101992818783194	decoded_dev_normalized_cer	macro_cell	0.096367	0.286279	0.096367	0.286279	0.079569	0.253559	0.079569	0.253559	5	1	1190
+14210	0.0427422970533371	0.4840303489675858	decoded_dev_normalized_cer	macro_cell	0.092797	0.267644	0.092797	0.267644	0.077902	0.236271	0.077902	0.236271	5	1	1190
+21315	0.007264204788953066	0.578342515937734	decoded_dev_normalized_cer	macro_cell	0.096725	0.267729	0.096725	0.267729	0.082294	0.238418	0.082294	0.238418	5	1	1190
+28420	0.0392894484102726	0.6794542931395289	decoded_dev_normalized_cer	macro_cell	0.094822	0.264003	0.094822	0.264003	0.083255	0.237175	0.083255	0.237175	5	1	1190
+35524	0.033548079431056976	0.65875939322298	decoded_dev_normalized_cer	macro_cell	0.101889	0.288365	0.101889	0.288365	0.087843	0.257401	0.087843	0.257401	5	1	1190

evaluation_results.json

@@ -0,0 +1,176 @@
+{
+  "aggregate": {
+    "cer_normalized_micro": 0.0792860273317662,
+    "character_errors": 3980,
+    "corpora": 5,
+    "reference_characters": 50198,
+    "reference_words": 8748,
+    "utterances": 1189,
+    "wer_normalized_micro": 0.2293095564700503,
+    "word_errors": 2006
+  },
+  "by_corpus": [
+    {
+      "cer_normalized": "0.08096",
+      "cer_raw": "0.08096",
+      "char_deletions": "682",
+      "char_deletions_raw": "682",
+      "char_insertions": "482",
+      "char_insertions_raw": "482",
+      "char_substitutions": "621",
+      "char_substitutions_raw": "621",
+      "eval_corpus": "ILRDF",
+      "eval_language": "saisiyat",
+      "eval_utterances": "610",
+      "reference_characters": "22048",
+      "reference_characters_raw": "22048",
+      "reference_words": "3766",
+      "reference_words_raw": "3766",
+      "run_id": "xlsr53_language_pooled_train-saisiyat_eval-saisiyat-all-corpora_dataset_v1_split_v2_formosan_safe_v1_seed13",
+      "wer_normalized": "0.21163",
+      "wer_raw": "0.21163",
+      "word_deletions": "108",
+      "word_deletions_raw": "108",
+      "word_insertions": "77",
+      "word_insertions_raw": "77",
+      "word_substitutions": "612",
+      "word_substitutions_raw": "612"
+    },
+    {
+      "cer_normalized": "0.283092",
+      "cer_raw": "0.283092",
+      "char_deletions": "542",
+      "char_deletions_raw": "542",
+      "char_insertions": "304",
+      "char_insertions_raw": "304",
+      "char_substitutions": "403",
+      "char_substitutions_raw": "403",
+      "eval_corpus": "NTU",
+      "eval_language": "saisiyat",
+      "eval_utterances": "104",
+      "reference_characters": "4412",
+      "reference_characters_raw": "4412",
+      "reference_words": "877",
+      "reference_words_raw": "877",
+      "run_id": "xlsr53_language_pooled_train-saisiyat_eval-saisiyat-all-corpora_dataset_v1_split_v2_formosan_safe_v1_seed13",
+      "wer_normalized": "0.651083",
+      "wer_raw": "0.651083",
+      "word_deletions": "89",
+      "word_deletions_raw": "89",
+      "word_insertions": "40",
+      "word_insertions_raw": "40",
+      "word_substitutions": "442",
+      "word_substitutions_raw": "442"
+    },
+    {
+      "cer_normalized": "0.028431",
+      "cer_raw": "0.028431",
+      "char_deletions": "18",
+      "char_deletions_raw": "18",
+      "char_insertions": "10",
+      "char_insertions_raw": "10",
+      "char_substitutions": "30",
+      "char_substitutions_raw": "30",
+      "eval_corpus": "ePark1",
+      "eval_language": "saisiyat",
+      "eval_utterances": "75",
+      "reference_characters": "2040",
+      "reference_characters_raw": "2040",
+      "reference_words": "343",
+      "reference_words_raw": "343",
+      "run_id": "xlsr53_language_pooled_train-saisiyat_eval-saisiyat-all-corpora_dataset_v1_split_v2_formosan_safe_v1_seed13",
+      "wer_normalized": "0.151603",
+      "wer_raw": "0.151603",
+      "word_deletions": "6",
+      "word_deletions_raw": "6",
+      "word_insertions": "2",
+      "word_insertions_raw": "2",
+      "word_substitutions": "44",
+      "word_substitutions_raw": "44"
+    },
+    {
+      "cer_normalized": "0.02624",
+      "cer_raw": "0.02624",
+      "char_deletions": "77",
+      "char_deletions_raw": "77",
+      "char_insertions": "27",
+      "char_insertions_raw": "27",
+      "char_substitutions": "24",
+      "char_substitutions_raw": "24",
+      "eval_corpus": "ePark2",
+      "eval_language": "saisiyat",
+      "eval_utterances": "106",
+      "reference_characters": "4878",
+      "reference_characters_raw": "4878",
+      "reference_words": "824",
+      "reference_words_raw": "824",
+      "run_id": "xlsr53_language_pooled_train-saisiyat_eval-saisiyat-all-corpora_dataset_v1_split_v2_formosan_safe_v1_seed13",
+      "wer_normalized": "0.106796",
+      "wer_raw": "0.106796",
+      "word_deletions": "16",
+      "word_deletions_raw": "16",
+      "word_insertions": "8",
+      "word_insertions_raw": "8",
+      "word_substitutions": "64",
+      "word_substitutions_raw": "64"
+    },
+    {
+      "cer_normalized": "0.045184",
+      "cer_raw": "0.045184",
+      "char_deletions": "371",
+      "char_deletions_raw": "371",
+      "char_insertions": "245",
+      "char_insertions_raw": "245",
+      "char_substitutions": "144",
+      "char_substitutions_raw": "144",
+      "eval_corpus": "ePark3",
+      "eval_language": "saisiyat",
+      "eval_utterances": "294",
+      "reference_characters": "16820",
+      "reference_characters_raw": "16820",
+      "reference_words": "2938",
+      "reference_words_raw": "2938",
+      "run_id": "xlsr53_language_pooled_train-saisiyat_eval-saisiyat-all-corpora_dataset_v1_split_v2_formosan_safe_v1_seed13",
+      "wer_normalized": "0.169503",
+      "wer_raw": "0.169503",
+      "word_deletions": "82",
+      "word_deletions_raw": "82",
+      "word_insertions": "48",
+      "word_insertions_raw": "48",
+      "word_substitutions": "368",
+      "word_substitutions_raw": "368"
+    }
+  ],
+  "condition": "language_pooled",
+  "dataset_version": "dataset_v1",
+  "iso_639_3": "xsy",
+  "language": "Saisiyat",
+  "language_id": "saisiyat",
+  "model_family": "xlsr53",
+  "normalization_version": "formosan_safe_v1",
+  "readiness": {
+    "actual_train_rows": 9473,
+    "best_selection_cer": 0.09279681756460974,
+    "best_selection_wer": 0.2676440207226045,
+    "best_step": 14210,
+    "condition": "language_pooled",
+    "dominant_hypothesis_rate": 0.001682085786375105,
+    "estimated_effective_epochs": 30.000211126359126,
+    "failures": 0,
+    "generated_at": "2026-06-23T08:17:51+00:00",
+    "intended_train_rows": 9473,
+    "manifest_dir": "/projects/prudlab/tacl_formosan_asr/experiments/manifests/gate4_xlsr_supervised_matrix/xlsr53_language_pooled_train-saisiyat_eval-saisiyat-all-corpora_dataset_v1_split_v2_formosan_safe_v1_seed13",
+    "nonempty_hypothesis_rate": 0.9991589571068125,
+    "prediction_rows": 1189,
+    "result_dir": "/scratch/scheppat/projects/tacl_formosan_asr/runs_xlsr_final_v2/xlsr53_language_pooled_train-saisiyat_eval-saisiyat-all-corpora_dataset_v1_split_v2_formosan_safe_v1_seed13",
+    "result_label": "final",
+    "run_id": "xlsr53_language_pooled_train-saisiyat_eval-saisiyat-all-corpora_dataset_v1_split_v2_formosan_safe_v1_seed13",
+    "status": "pass",
+    "steps": 35524,
+    "test_rows": 1189,
+    "train_batch_size": 8,
+    "warnings": 0
+  },
+  "seed": 13,
+  "split_version": "split_v2"
+}

metrics_by_corpus.tsv

@@ -0,0 +1,6 @@
+run_id	eval_language	eval_corpus	eval_utterances	wer_raw	cer_raw	wer_normalized	cer_normalized	word_substitutions	word_deletions	word_insertions	reference_words	char_substitutions	char_deletions	char_insertions	reference_characters	word_substitutions_raw	word_deletions_raw	word_insertions_raw	reference_words_raw	char_substitutions_raw	char_deletions_raw	char_insertions_raw	reference_characters_raw
+xlsr53_language_pooled_train-saisiyat_eval-saisiyat-all-corpora_dataset_v1_split_v2_formosan_safe_v1_seed13	saisiyat	ILRDF	610	0.21163	0.08096	0.21163	0.08096	612	108	77	3766	621	682	482	22048	612	108	77	3766	621	682	482	22048
+xlsr53_language_pooled_train-saisiyat_eval-saisiyat-all-corpora_dataset_v1_split_v2_formosan_safe_v1_seed13	saisiyat	NTU	104	0.651083	0.283092	0.651083	0.283092	442	89	40	877	403	542	304	4412	442	89	40	877	403	542	304	4412
+xlsr53_language_pooled_train-saisiyat_eval-saisiyat-all-corpora_dataset_v1_split_v2_formosan_safe_v1_seed13	saisiyat	ePark1	75	0.151603	0.028431	0.151603	0.028431	44	6	2	343	30	18	10	2040	44	6	2	343	30	18	10	2040
+xlsr53_language_pooled_train-saisiyat_eval-saisiyat-all-corpora_dataset_v1_split_v2_formosan_safe_v1_seed13	saisiyat	ePark2	106	0.106796	0.02624	0.106796	0.02624	64	16	8	824	24	77	27	4878	64	16	8	824	24	77	27	4878
+xlsr53_language_pooled_train-saisiyat_eval-saisiyat-all-corpora_dataset_v1_split_v2_formosan_safe_v1_seed13	saisiyat	ePark3	294	0.169503	0.045184	0.169503	0.045184	368	82	48	2938	144	371	245	16820	368	82	48	2938	144	371	245	16820

model.safetensors

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:c1fb355becaa03b31effb5bde521abdaf9f1c3e4860a44ceb936b5a5db4c8031
+size 1261938588

preprocessor_config.json

@@ -0,0 +1,10 @@
+{
+  "do_normalize": true,
+  "feature_extractor_type": "Wav2Vec2FeatureExtractor",
+  "feature_size": 1,
+  "padding_side": "right",
+  "padding_value": 0.0,
+  "processor_class": "Wav2Vec2Processor",
+  "return_attention_mask": true,
+  "sampling_rate": 16000
+}

readiness_summary.json

@@ -0,0 +1,23 @@
+{
+  "actual_train_rows": 9473,
+  "best_selection_cer": 0.09279681756460974,
+  "best_selection_wer": 0.2676440207226045,
+  "best_step": 14210,
+  "condition": "language_pooled",
+  "dominant_hypothesis_rate": 0.001682085786375105,
+  "estimated_effective_epochs": 30.000211126359126,
+  "failures": 0,
+  "generated_at": "2026-06-23T08:17:51+00:00",
+  "intended_train_rows": 9473,
+  "manifest_dir": "/projects/prudlab/tacl_formosan_asr/experiments/manifests/gate4_xlsr_supervised_matrix/xlsr53_language_pooled_train-saisiyat_eval-saisiyat-all-corpora_dataset_v1_split_v2_formosan_safe_v1_seed13",
+  "nonempty_hypothesis_rate": 0.9991589571068125,
+  "prediction_rows": 1189,
+  "result_dir": "/scratch/scheppat/projects/tacl_formosan_asr/runs_xlsr_final_v2/xlsr53_language_pooled_train-saisiyat_eval-saisiyat-all-corpora_dataset_v1_split_v2_formosan_safe_v1_seed13",
+  "result_label": "final",
+  "run_id": "xlsr53_language_pooled_train-saisiyat_eval-saisiyat-all-corpora_dataset_v1_split_v2_formosan_safe_v1_seed13",
+  "status": "pass",
+  "steps": 35524,
+  "test_rows": 1189,
+  "train_batch_size": 8,
+  "warnings": 0
+}

release_manifest.json

@@ -0,0 +1,97 @@
+{
+  "aggregate_metrics": {
+    "cer_normalized_micro": 0.0792860273317662,
+    "character_errors": 3980,
+    "corpora": 5,
+    "reference_characters": 50198,
+    "reference_words": 8748,
+    "utterances": 1189,
+    "wer_normalized_micro": 0.2293095564700503,
+    "word_errors": 2006
+  },
+  "base_model": "facebook/wav2vec2-large-xlsr-53",
+  "condition": "language_pooled",
+  "files": [
+    {
+      "bytes": 50,
+      "name": ".gitattributes",
+      "sha256": "ee05350c7fda0af4da116020dd7b976d731de3beed4f5e717d5bcc95ead24550"
+    },
+    {
+      "bytes": 4182,
+      "name": "README.md",
+      "sha256": "c23d8719005fe9d89a14e81f6cc09feb95edce48ca6ed13f6c5085575c34498a"
+    },
+    {
+      "bytes": 30,
+      "name": "added_tokens.json",
+      "sha256": "2f1da08637d8de6300081239c996342b9c89aa9e1104f167b980fea403358a1f"
+    },
+    {
+      "bytes": 2335,
+      "name": "config.json",
+      "sha256": "d8d236d9d520a37e96eb9812b3bf1e4a714debb58123b7b9a3ab02bb569b2d50"
+    },
+    {
+      "bytes": 1188,
+      "name": "dev_selection_history.tsv",
+      "sha256": "02e920c678eeaf49abc585a66ad0c6221af59bb620ed8dc4dc935fd30c1902ea"
+    },
+    {
+      "bytes": 6351,
+      "name": "evaluation_results.json",
+      "sha256": "49c6f688d1b08c129a56a1d3097ecfaae231df20b9efd3e8e0b603268d7b6b88"
+    },
+    {
+      "bytes": 1514,
+      "name": "metrics_by_corpus.tsv",
+      "sha256": "929e84dcd7422a42e1f2c8db9dd4ca9006f3dfa4610525074f0a1f02efbc2066"
+    },
+    {
+      "bytes": 1261938588,
+      "name": "model.safetensors",
+      "sha256": "c1fb355becaa03b31effb5bde521abdaf9f1c3e4860a44ceb936b5a5db4c8031"
+    },
+    {
+      "bytes": 256,
+      "name": "preprocessor_config.json",
+      "sha256": "ffc72f475cd79680d5d659fbaefc691cdbcee5c9703a15faf708a5e91452a8d8"
+    },
+    {
+      "bytes": 1111,
+      "name": "readiness_summary.json",
+      "sha256": "a6d725e7ef8d7547f6ec9dee17134c414144d6724d924dbd27585e7ee08f6834"
+    },
+    {
+      "bytes": 96,
+      "name": "special_tokens_map.json",
+      "sha256": "9046da57c270c8e74d0f38832b4adce269c9d914ef21d2a0925e7772152dd793"
+    },
+    {
+      "bytes": 1096,
+      "name": "tokenizer_config.json",
+      "sha256": "5ec9d1d8004284dc7081e99c75280e2f7685491ba7f9c071b41e4e74cfedc63b"
+    },
+    {
+      "bytes": 5969,
+      "name": "training_config.json",
+      "sha256": "64f9909ec45fd93f448a8488355a5ce7b30329dab5d3040e017b92162e97cc1a"
+    },
+    {
+      "bytes": 1420,
+      "name": "training_summary.json",
+      "sha256": "75de33dce242b11915676883486a1246aad89bab518f56b2e8ca1a39b2a01c1f"
+    },
+    {
+      "bytes": 342,
+      "name": "vocab.json",
+      "sha256": "b99f83ad72973910f493aeee8abc9822fd1f1f7a7d957cdbca51ce290cdb700e"
+    }
+  ],
+  "iso_639_3": "xsy",
+  "language": "Saisiyat",
+  "model_family": "xlsr53",
+  "readiness_status": "pass",
+  "repo_id": "FormosanBank/formosan-asr-saisiyat",
+  "source_run": "/scratch/scheppat/projects/tacl_formosan_asr/runs_xlsr_final_v2/xlsr53_language_pooled_train-saisiyat_eval-saisiyat-all-corpora_dataset_v1_split_v2_formosan_safe_v1_seed13"
+}

special_tokens_map.json

@@ -0,0 +1,6 @@
+{
+  "bos_token": "<s>",
+  "eos_token": "</s>",
+  "pad_token": "<pad>",
+  "unk_token": "<unk>"
+}

tokenizer_config.json

@@ -0,0 +1,48 @@
+{
+  "added_tokens_decoder": {
+    "29": {
+      "content": "<unk>",
+      "lstrip": true,
+      "normalized": false,
+      "rstrip": true,
+      "single_word": false,
+      "special": false
+    },
+    "30": {
+      "content": "<pad>",
+      "lstrip": true,
+      "normalized": false,
+      "rstrip": true,
+      "single_word": false,
+      "special": false
+    },
+    "31": {
+      "content": "<s>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": true
+    },
+    "32": {
+      "content": "</s>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": true
+    }
+  },
+  "bos_token": "<s>",
+  "clean_up_tokenization_spaces": false,
+  "do_lower_case": false,
+  "eos_token": "</s>",
+  "model_max_length": 1000000000000000019884624838656,
+  "pad_token": "<pad>",
+  "processor_class": "Wav2Vec2Processor",
+  "replace_word_delimiter_char": " ",
+  "target_lang": null,
+  "tokenizer_class": "Wav2Vec2CTCTokenizer",
+  "unk_token": "<unk>",
+  "word_delimiter_token": "|"
+}

training_config.json

@@ -0,0 +1,187 @@
+{
+  "code_git_commit": "70c74ec04f113a8aab9376ee0e05f62f83f0c4b3",
+  "code_git_dirty": "unknown",
+  "condition": "language_pooled",
+  "cuda_visible_devices": "0",
+  "dataset_version": "dataset_v1",
+  "dev_corpus": "",
+  "dev_hours": 2.024895,
+  "dev_language": "saisiyat",
+  "dev_rows": 1190,
+  "dev_scope": "language_pooled",
+  "device": "cuda",
+  "effective_eval_every_steps": 7105,
+  "effective_max_steps": 35524,
+  "eval_batch_size": 8,
+  "eval_corpus": "",
+  "eval_every_steps": 7105,
+  "eval_language": "saisiyat",
+  "eval_scope": "language_pooled",
+  "generated_at": "2026-06-23T04:18:52+00:00",
+  "hostname": "g018",
+  "max_steps": 1000,
+  "model_family": "xlsr53",
+  "model_id": "facebook/wav2vec2-large-xlsr-53",
+  "model_revision": "c3f9d884181a224a6ac87bf8885c84d1cff3384f",
+  "normalization_version": "formosan_safe_v1",
+  "package_freeze": [
+    "absl-py==2.4.0",
+    "accelerate==1.1.1",
+    "aiohappyeyeballs==2.6.2",
+    "aiohttp==3.14.1",
+    "aiosignal==1.4.0",
+    "async-timeout==5.0.1",
+    "attrs==26.1.0",
+    "audioread==3.1.0",
+    "certifi==2026.6.17",
+    "cffi==2.0.0",
+    "charset-normalizer==3.4.7",
+    "click==8.4.1",
+    "datasets==3.1.0",
+    "decorator==5.3.1",
+    "dill==0.3.8",
+    "evaluate==0.4.3",
+    "filelock==3.29.0",
+    "frozenlist==1.8.0",
+    "fsspec==2024.9.0",
+    "grpcio==1.81.1",
+    "hf-xet==1.5.1",
+    "huggingface_hub==0.36.2",
+    "idna==3.18",
+    "Jinja2==3.1.6",
+    "jiwer==3.0.5",
+    "joblib==1.5.3",
+    "lazy-loader==0.5",
+    "librosa==0.10.2.post1",
+    "llvmlite==0.47.0",
+    "Markdown==3.10.2",
+    "MarkupSafe==3.0.3",
+    "mpmath==1.3.0",
+    "msgpack==1.2.1",
+    "multidict==6.7.1",
+    "multiprocess==0.70.16",
+    "networkx==3.4.2",
+    "numba==0.65.1",
+    "numpy==2.2.6",
+    "nvidia-cublas-cu12==12.4.5.8",
+    "nvidia-cuda-cupti-cu12==12.4.127",
+    "nvidia-cuda-nvrtc-cu12==12.4.127",
+    "nvidia-cuda-runtime-cu12==12.4.127",
+    "nvidia-cudnn-cu12==9.1.0.70",
+    "nvidia-cufft-cu12==11.2.1.3",
+    "nvidia-curand-cu12==10.3.5.147",
+    "nvidia-cusolver-cu12==11.6.1.9",
+    "nvidia-cusparse-cu12==12.3.1.170",
+    "nvidia-nccl-cu12==2.21.5",
+    "nvidia-nvjitlink-cu12==12.4.127",
+    "nvidia-nvtx-cu12==12.4.127",
+    "packaging @ file:///home/conda/feedstock_root/build_artifacts/bld/rattler-build_packaging_1777103621/work",
+    "pandas==2.3.3",
+    "platformdirs==4.10.0",
+    "pooch==1.9.0",
+    "propcache==0.5.2",
+    "protobuf==7.35.1",
+    "psutil==7.2.2",
+    "pyarrow==24.0.0",
+    "pycparser==3.0",
+    "python-dateutil==2.9.0.post0",
+    "pytz==2026.2",
+    "PyYAML==6.0.3",
+    "RapidFuzz==3.14.5",
+    "regex==2026.5.9",
+    "requests==2.34.2",
+    "safetensors==0.8.0",
+    "scikit-learn==1.7.2",
+    "scipy==1.15.3",
+    "sentencepiece==0.2.0",
+    "six==1.17.0",
+    "soundfile==0.12.1",
+    "soxr==1.1.0",
+    "sympy==1.13.1",
+    "tensorboard==2.18.0",
+    "tensorboard-data-server==0.7.2",
+    "threadpoolctl==3.6.0",
+    "tokenizers==0.20.3",
+    "torch==2.5.1+cu124",
+    "torchaudio==2.5.1+cu124",
+    "tqdm==4.68.3",
+    "transformers==4.46.3",
+    "triton==3.1.0",
+    "typing_extensions==4.15.0",
+    "tzdata==2026.2",
+    "urllib3==2.7.0",
+    "Werkzeug==3.1.8",
+    "xxhash==3.7.0",
+    "yarl==1.24.2"
+  ],
+  "package_freeze_count": 88,
+  "python_argv": [
+    "/projects/prudlab/tacl_formosan_asr/code/tacl_pipeline/src/tacl_formosan_asr/train_ctc_asr.py",
+    "--source-root",
+    "/projects/prudlab/hunter_formosan_audio",
+    "--run-dir",
+    "/projects/prudlab/tacl_formosan_asr/experiments/manifests/gate4_xlsr_supervised_matrix/xlsr53_language_pooled_train-saisiyat_eval-saisiyat-all-corpora_dataset_v1_split_v2_formosan_safe_v1_seed13",
+    "--vocab-json",
+    "/projects/prudlab/tacl_formosan_asr/vocabularies/ctc_dataset_v1_split_v2_formosan_safe_v1/languages/saisiyat/vocab.json",
+    "--out-dir",
+    "/scratch/scheppat/projects/tacl_formosan_asr/runs_xlsr_final_v2/xlsr53_language_pooled_train-saisiyat_eval-saisiyat-all-corpora_dataset_v1_split_v2_formosan_safe_v1_seed13",
+    "--run-id",
+    "xlsr53_language_pooled_train-saisiyat_eval-saisiyat-all-corpora_dataset_v1_split_v2_formosan_safe_v1_seed13",
+    "--model-id",
+    "facebook/wav2vec2-large-xlsr-53",
+    "--model-revision",
+    "c3f9d884181a224a6ac87bf8885c84d1cff3384f",
+    "--seed",
+    "13",
+    "--max-steps",
+    "1000",
+    "--train-epochs",
+    "30",
+    "--train-batch-size",
+    "8",
+    "--eval-batch-size",
+    "8",
+    "--learning-rate",
+    "1e-4",
+    "--eval-every-steps",
+    "0",
+    "--max-eval-batches",
+    "0",
+    "--max-train-rows",
+    "0",
+    "--max-dev-rows",
+    "0",
+    "--max-test-rows",
+    "0",
+    "--save-best-model"
+  ],
+  "python_executable": "/scratch/scheppat/projects/tacl_formosan_asr/conda/tacl_asr_py310/bin/python",
+  "python_version": "3.10.20 | packaged by conda-forge | (main, Jun 11 2026, 03:31:56) [GCC 14.3.0]",
+  "requested_eval_every_steps": 0,
+  "run_dir": "/projects/prudlab/tacl_formosan_asr/experiments/manifests/gate4_xlsr_supervised_matrix/xlsr53_language_pooled_train-saisiyat_eval-saisiyat-all-corpora_dataset_v1_split_v2_formosan_safe_v1_seed13",
+  "run_id": "xlsr53_language_pooled_train-saisiyat_eval-saisiyat-all-corpora_dataset_v1_split_v2_formosan_safe_v1_seed13",
+  "seed": 13,
+  "slurm_job_id": "2630081",
+  "slurm_job_name": "tacl_ctc_registry_array",
+  "split_version": "split_v2",
+  "status": "planned",
+  "tacl_code_version_file": "70c74ec04f113a8aab9376ee0e05f62f83f0c4b3",
+  "test_hours": 2.024643,
+  "test_rows": 1189,
+  "train_batch_size": 8,
+  "train_corpus": "",
+  "train_corpus_counts": {
+    "ILRDF": 4870,
+    "NTU": 826,
+    "ePark1": 594,
+    "ePark2": 838,
+    "ePark3": 2345
+  },
+  "train_epochs": 30.0,
+  "train_hours": 16.209974,
+  "train_language": "saisiyat",
+  "train_rows": 9473,
+  "train_scope": "language_pooled",
+  "vocab_json": "/projects/prudlab/tacl_formosan_asr/vocabularies/ctc_dataset_v1_split_v2_formosan_safe_v1/languages/saisiyat/vocab.json",
+  "vocab_scope": "language"
+}

training_summary.json

@@ -0,0 +1,31 @@
+{
+  "best_dev_loss": 0.4840303489675858,
+  "best_selection_cer": 0.09279681756460974,
+  "best_selection_metrics": {
+    "dev_cells": 5,
+    "dev_languages": 1,
+    "dev_rows": 1190,
+    "macro_cell_cer": 0.09279681756460974,
+    "macro_cell_wer": 0.2676440207226045,
+    "macro_language_cer": 0.07790196078431373,
+    "macro_language_wer": 0.23627118644067796,
+    "micro_cer": 0.07790196078431373,
+    "micro_wer": 0.23627118644067796,
+    "selection_cer": 0.09279681756460974,
+    "selection_metric": "decoded_dev_normalized_cer",
+    "selection_scope": "macro_cell",
+    "selection_wer": 0.2676440207226045
+  },
+  "best_selection_wer": 0.2676440207226045,
+  "best_step": 14210,
+  "checkpoint_path": "/scratch/scheppat/projects/tacl_formosan_asr/runs_xlsr_final_v2/xlsr53_language_pooled_train-saisiyat_eval-saisiyat-all-corpora_dataset_v1_split_v2_formosan_safe_v1_seed13/best_model",
+  "end_train_loss": 0.033548079431056976,
+  "generated_at": "2026-06-23T08:17:48+00:00",
+  "max_steps": 1000,
+  "out_dir": "/scratch/scheppat/projects/tacl_formosan_asr/runs_xlsr_final_v2/xlsr53_language_pooled_train-saisiyat_eval-saisiyat-all-corpora_dataset_v1_split_v2_formosan_safe_v1_seed13",
+  "prediction_rows": 1189,
+  "run_id": "xlsr53_language_pooled_train-saisiyat_eval-saisiyat-all-corpora_dataset_v1_split_v2_formosan_safe_v1_seed13",
+  "start_train_loss": 21.27985954284668,
+  "steps": 35524,
+  "train_epochs": 30.0
+}

vocab.json

@@ -0,0 +1,33 @@
+{
+  "'": 0,
+  "-": 1,
+  ":": 2,
+  "<pad>": 30,
+  "<unk>": 29,
+  "^": 3,
+  "a": 4,
+  "b": 5,
+  "c": 6,
+  "d": 7,
+  "e": 8,
+  "f": 9,
+  "g": 10,
+  "h": 11,
+  "i": 12,
+  "k": 13,
+  "l": 14,
+  "m": 15,
+  "n": 16,
+  "o": 17,
+  "p": 18,
+  "q": 19,
+  "r": 20,
+  "s": 21,
+  "t": 22,
+  "u": 23,
+  "w": 24,
+  "x": 25,
+  "y": 26,
+  "z": 27,
+  "|": 28
+}