Update README.md

README.md

@@ -1,199 +1,258 @@
 ---
 library_name: transformers
-tags: []
+tags:
+- translation
+- Tibetan
+- Buddhism
+- dharma
+license: mit
+language:
+- bo
+- en
+metrics:
+- bleu
+- ter
+- chrf
+base_model:
+- google-t5/t5-base
+pipeline_tag: translation
 ---
 
-# Model Card for Model ID
-
-<!-- Provide a quick summary of what the model is/does. -->
-
+# Model Card for mlotsawa-ground-small
 
+This model is a transformers machine translation model for translating Tibetan Buddhist texts to English, produced as part of the larger [MLotsawa project](https://github.com/billingsmoore/MLotsawa)
 
 ## Model Details
 
 ### Model Description
 
-<!-- Provide a longer summary of what this model is. -->
-
-This is the model card of a 🤗 transformers model that has been pushed on the Hub. This model card has been automatically generated.
+This model is a finetuned T5 model (base size) with 223 million parameters. It is intended for translation of Tibetan Buddhist texts into English. It expects input in Uchen script. 
+This model uses the **[getok](https://huggingface.co/billingsmoore/getok-v0)** tokenizer.
+Details on the training data and procedure can be found below.
 
-- **Developed by:** [More Information Needed]
-- **Funded by [optional]:** [More Information Needed]
-- **Shared by [optional]:** [More Information Needed]
-- **Model type:** [More Information Needed]
-- **Language(s) (NLP):** [More Information Needed]
-- **License:** [More Information Needed]
-- **Finetuned from model [optional]:** [More Information Needed]
+This model is a *ground* model in that, while its performance is reasonably good, it is intended to be used as a base for further finetuning on either a larger corpus or a tradition-specific (i.e. Dzogchen) corpus for improved translation quality.
 
-### Model Sources [optional]
+- **Developed by:** billingsmoore
+- **Model type:** translation
+- **Languages:** Tibetan, English
+- **License:** MIT
+- **Finetuned from model:** google-t5/t5-base
 
-<!-- Provide the basic links for the model. -->
+### Model Sources
 
-- **Repository:** [More Information Needed]
-- **Paper [optional]:** [More Information Needed]
-- **Demo [optional]:** [More Information Needed]
+- **Repository:** [MLotsawa on GitHub](https://github.com/billingsmoore/MLotsawa)
 
 ## Uses
 
-<!-- Address questions around how the model is intended to be used, including the foreseeable users of the model and those affected by the model. -->
+This model may be used directly for translation, or further finetuned for improved performance.
 
 ### Direct Use
 
-<!-- This section is for the model use without fine-tuning or plugging into a larger ecosystem/app. -->
-
-[More Information Needed]
-
-### Downstream Use [optional]
-
-<!-- This section is for the model use when fine-tuned for a task, or when plugged into a larger ecosystem/app -->
-
-[More Information Needed]
-
-### Out-of-Scope Use
-
-<!-- This section addresses misuse, malicious use, and uses that the model will not work well for. -->
-
-[More Information Needed]
-
-## Bias, Risks, and Limitations
-
-<!-- This section is meant to convey both technical and sociotechnical limitations. -->
-
-[More Information Needed]
-
-### Recommendations
-
-<!-- This section is meant to convey recommendations with respect to the bias, risk, and technical limitations. -->
-
-Users (both direct and downstream) should be made aware of the risks, biases and limitations of the model. More information needed for further recommendations.
+This model can be used directly for translation using a transformers pipeline as in the code block below.
 
-## How to Get Started with the Model
+```python
+from transformers import pipeline
 
-Use the code below to get started with the model.
+pipe = pipeline('translation', 'billingsmoore/mlotsawa-ground-base', device='cpu') # select a device of your choice (i.e. 'cuda:0')
 
-[More Information Needed]
+input = ["ཁྱེད་ལ་བསྟོད་ཅིང་གསོལ་བ་བཏབ་པའི་མཐུས༔",
+"བདག་གི་ཚེ་བསོད་དཔལ་འབྱོར་རྒྱས་པ་དང་༔",
+"འཇིགས་པ་བཅུ་དྲུག་རྐྱེན་ངན་བར་ཆད་སོལ༔"]
 
-## Training Details
-
-### Training Data
+output = pipe(input)
 
-<!-- This should link to a Dataset Card, perhaps with a short stub of information on what the training data is all about as well as documentation related to data pre-processing or additional filtering. -->
+translation = [elt['translation_text'] for elt in output]
 
-[More Information Needed]
-
-### Training Procedure
+print(translation)
+```
 
-<!-- This relates heavily to the Technical Specifications. Content here should link to that section when it is relevant to the training procedure. -->
+The code above will produce the following output.
 
-#### Preprocessing [optional]
+>['Through the power of praising and praying to you', 'Increase my lifespan merit and prosperity', 'Remove the sixteen fears and obstacles of adversity.']
 
-[More Information Needed]
+Alternatively the model can be used with a graphical user interface following [the instructions found here.](https://pypi.org/project/mlotsawa/)
 
+### Downstream Use
 
-#### Training Hyperparameters
+The performance of this model can be improved with additional finetuning. You might finetune using a larger dataset for better general performance or finetune on a specific set of material for improved performance on that subset (i.e. Dzogchen texts).
 
-- **Training regime:** [More Information Needed] <!--fp32, fp16 mixed precision, bf16 mixed precision, bf16 non-mixed precision, fp16 non-mixed precision, fp8 mixed precision -->
+The model can be finetuned following the recipe below.
 
-#### Speeds, Sizes, Times [optional]
-
-<!-- This section provides information about throughput, start/end time, checkpoint size if relevant, etc. -->
-
-[More Information Needed]
-
-## Evaluation
+```python
+# Load Your Data
+from datasets import load_dataset
 
-<!-- This section describes the evaluation protocols and provides the results. -->
+dataset = load_dataset(<your dataset>)
 
-### Testing Data, Factors & Metrics
+# Load the Model and Tokenizer
+from transformers import AutoTokenizer, DataCollatorForSeq2Seq, AutoModelForSeq2SeqLM
 
-#### Testing Data
+model = AutoModelForSeq2SeqLM.from_pretrained("billingsmoore/mlotsawa-ground-base", device_map="cuda:0") # this line assumes you want to use a single CUDA enabled gpu
+tokenizer = AutoTokenizer.from_pretrained('billingsmoore/mlotsawa-ground-base')
+data_collator = DataCollatorForSeq2Seq(tokenizer=tokenizer, model=model)
 
-<!-- This should link to a Dataset Card if possible. -->
+# Preprocess the Data
+def translation_preprocess_function(examples):
 
-[More Information Needed]
+    # Prepare translation inputs and targets
+    translation_inputs = ['Translate Tibetan to English: ' + example for example in examples['bo']]
+    translation_targets = [example for example in examples['en']]
+    
+    # Tokenize translation inputs and targets
+    translation_model_inputs = tokenizer(translation_inputs, text_target=translation_targets, 
+                                         max_length=256, truncation=True, padding="max_length")
+    
+    
+    return translation_model_inputs
 
-#### Factors
+tokenized_dataset = dataset.map(translation_preprocess_function, batched=True)
 
-<!-- These are the things the evaluation is disaggregating by, e.g., subpopulations or domains. -->
+# Define Evaluation Metrics
+import numpy as np
+import evaluate
 
-[More Information Needed]
+# Load BLEU and CHRF metrics
+bleu_metric = evaluate.load("sacrebleu")
+chrf_metric = evaluate.load("chrf")
+ter_metric = evaluate.load("ter")
 
-#### Metrics
+def postprocess_text(preds, labels):
+    preds = [pred.strip() for pred in preds]
+    labels = [[label.strip()] for label in labels]
 
-<!-- These are the evaluation metrics being used, ideally with a description of why. -->
+    return preds, labels
 
-[More Information Needed]
+def compute_metrics(eval_preds):
+    preds, labels = eval_preds
+    if isinstance(preds, tuple):
+        preds = preds[0]
+    
+    # Decode predictions and labels
+    preds = np.where(preds != -100, preds, tokenizer.pad_token_id)
+    decoded_preds = tokenizer.batch_decode(preds, skip_special_tokens=True)
+    labels = np.where(labels != -100, labels, tokenizer.pad_token_id)
+    decoded_labels = tokenizer.batch_decode(labels, skip_special_tokens=True)
 
-### Results
+    # Postprocess text
+    decoded_preds, decoded_labels = postprocess_text(decoded_preds, decoded_labels)
 
-[More Information Needed]
+    # Compute BLEU score
+    bleu_result = bleu_metric.compute(predictions=decoded_preds, references=decoded_labels)
+    bleu_score = bleu_result["score"]
 
-#### Summary
+    # Compute CHRF score
+    chrf_result = chrf_metric.compute(predictions=decoded_preds, references=decoded_labels)
+    chrf_score = chrf_result["score"]
 
+    # Compute TER score
+    ter_result = ter_metric.compute(predictions=decoded_preds, references=decoded_labels)
+    ter_score = ter_result["score"]
 
+    # Return rounded results
+    metrics = {
+        "bleu": round(bleu_score, 4),
+        "chrf": round(chrf_score, 4),
+        "ter": round(ter_score, 4)
+    }
 
-## Model Examination [optional]
+    #print("Computed Metrics:", metrics)
 
-<!-- Relevant interpretability work for the model goes here -->
+    return metrics
 
-[More Information Needed]
+# Set Up Training Arguments and Optimizer
+from transformers import Seq2SeqTrainingArguments, Seq2SeqTrainer, Adafactor, EarlyStoppingCallback
+from accelerate import Accelerator
+
+accelerator = Accelerator()
+
+optimizer = Adafactor(
+    model.parameters(), 
+    scale_parameter=True, 
+    relative_step=False, 
+    warmup_init=False, 
+    lr=3e-4
+)
+
+model, optimizer = accelerator.prepare(model, optimizer)
+
+training_args = Seq2SeqTrainingArguments(
+    output_dir=f"output-dir", # select an output directory of your choice
+    auto_find_batch_size=True,
+    predict_with_generate=True,
+    fp16=False,
+    push_to_hub=False,
+    eval_strategy='epoch',
+    save_strategy='epoch',
+    num_train_epochs=100, # select your preferred number of training epochs
+    load_best_model_at_end=True,
+)
+
+trainer = Seq2SeqTrainer(
+    model=model,
+    args=training_args,
+    train_dataset=tokenized_dataset['train'],
+    eval_dataset=tokenized_dataset['dev'],
+    processing_class=tokenizer,
+    optimizers=(optimizer, None),
+    data_collator=data_collator,
+    compute_metrics=compute_metrics,
+    callbacks=[EarlyStoppingCallback()]
+)
+
+trainer.train()
+```
 
-## Environmental Impact
-
-<!-- Total emissions (in grams of CO2eq) and additional considerations, such as electricity usage, go here. Edit the suggested text below accordingly -->
-
-Carbon emissions can be estimated using the [Machine Learning Impact calculator](https://mlco2.github.io/impact#compute) presented in [Lacoste et al. (2019)](https://arxiv.org/abs/1910.09700).
-
-- **Hardware Type:** [More Information Needed]
-- **Hours used:** [More Information Needed]
-- **Cloud Provider:** [More Information Needed]
-- **Compute Region:** [More Information Needed]
-- **Carbon Emitted:** [More Information Needed]
+## Bias, Risks, and Limitations
 
-## Technical Specifications [optional]
+This model is intended for the translation of Buddhist texts. Because of the complexity and importance of this material, all translations should be treated as preliminary and should never be used without the input of an experienced human translator.
 
-### Model Architecture and Objective
+Additionally, this model was trained exclusively on Tibetan Buddhist material and should not be expected to perform well on other material (i.e. vernacular Tibetan).
 
-[More Information Needed]
+## Training Details
 
-### Compute Infrastructure
+### Training Data
 
-[More Information Needed]
+The training data for this model was 861,417 translation pairs from Buddhist texts. This data was collected from publically available material as well as material generously provided by Monlam AI and the Tibetan and Himalayan Library.
 
-#### Hardware
+### Training Procedure
 
-[More Information Needed]
+The model underwent continued pretraining as well as finetuning as described below.
 
-#### Software
+#### Pretraining
 
-[More Information Needed]
+The model was pretrained on the training data for one epoch with a learning rate of 3e-4.The pretraining objective remained the original span corruption denoising task, in which random spans of input tokens are masked and the model is trained to reconstruct the missing content. This pretraining allowed the model to adapt to the new tokenizer and to learn the linguistic and structural characteristics of the Tibetan Buddhist materials. 
 
-## Citation [optional]
+#### Finetuning
 
-<!-- If there is a paper or blog post introducing the model, the APA and Bibtex information for that should go in this section. -->
+The model was finetuned on the translation pairs for 50 epochs using the Adafactor optimizer and an initial learning rate of 3e-4.
 
-**BibTeX:**
+## Evaluation
 
-[More Information Needed]
+The model was evaluated on test data with BLEU, chrF, and TER. The results are shown below.
 
-**APA:**
+BLEU|chrF|TER
+----|----|---
+4.08|20.85|86.74
 
-[More Information Needed]
+These scores are exceptionally low, however actual translation results are relatively good. Sample translations are shown below.
 
-## Glossary [optional]
+From *Advice on Bending Mind Toward the Good* by Khenchen Ngawang Palzang | Translated by Joseph McClellan with editorial assistance from Ninjyed N.T., 2024.
 
-<!-- If relevant, include terms and calculations in this section that can help readers understand the model or model card. -->
+| **Original** | **Human Translation** | **Machine Translation** |
+|:-------------|:----------------------|:------------------------|
+| གྲུབ་བརྒྱའི་སྤྱི་མེས་པཎ་ཆེན་བི་མ་ལ། །<br>བསམ་བཞིན་སྤྲུལ་པའི་ཟློས་གར་ཉེར་བཟུང་བ། །<br>རྒྱལ་བའི་དབང་པོ་ཀློང་ཆེན་རབ་འབྱམས་པ། །<br>འདི་ཙམ་མ་ཡིན་ཚེ་རབས་གཏན་གྱི་སྐྱབས། ། | Grandsire of a hundred siddhas—great scholar, Vimalamitra,<br>And you who fully embraced the spectacle of intentional emanation,<br>Lord of conquerors, Longchen Rabjam—<br>You are my unfailing refuge; not just now, but in the concatenation of my lives. | Great paṇḍita Vimalamitra, forefather of hundreds of siddhas,<br>Manifesting in the form of a play,<br>Lord of the victorious ones, Longchen Rabjam,<br>Not just this but the constant refuge throughout all my lives, |
 
-[More Information Needed]
+From *Protection from All Fears A Prayer to Ārya Tārā from the Reality Ḍākinīs’ Secret Treasury (Chönyi Khandrö Sangdzö)* by Sera Khandro | Translated by Adam Pearcey, 2025.
 
-## More Information [optional]
+| **Original** | **Human Translation** | **Machine Translation** |
+|:-------------|:----------------------|:------------------------|
+|ཀ་དག་སྤྲོས་བྲལ་འོད་གསལ་རིག་པའི་དབྱིངས༔<br>ལྷུན་གྲུབ་སྣང་ཆ་མ་འགགས་སྒྱུ་འཕྲུལ་གར༔<br>ཐུགས་རྗེ་རྒྱལ་བ་ཀུན་གྱི་ཡུམ་གཅིག་མ༔<br>རྗེ་བཙུན་ཨཱརྱ་ཏཱ་རེ་ཚེ་སྦྱིན་དཔལ༔<br>གསོལ་བ་འདེབས་སོ་རླུང་སེམས་དབང་བསྡུས་ནས༔<br>ཚེ་དང་བསོད་ནམས་འཕེལ་བར་མཛད་དུ་གསོལ༔ | Out of the primordially pure unelaborate space of luminous awareness,<br>As the magical manifestation of unobstructed spontaneous presence,<br>Arises the compassionate one, the one and only mother of all victorious ones,<br>Noble Lady Ārya Tārā, glorious bestower of longevity,<br>To you I pray! Take control of my vital winds and mind,<br>And increase my lifespan and merit! | Within the space of awareness—primordial purity free of elaboration—<br>Illusory dance of spontaneously present appearances unceasing<br>Only mother of all the buddhas of compassion<br>Noble Ārya Tārā glorious Tārā<br>To you I pray: bringing the vāyu-mind under control<br>And increase our lifespan and merit.|
 
-[More Information Needed]
 
-## Model Card Authors [optional]
+## Model Card Authors
 
-[More Information Needed]
+billingsmoore
 
 ## Model Card Contact
 
-[More Information Needed]
+billingsmoore[at]gmail[dot]com