complete readme draft

README.md

@@ -5,38 +5,44 @@ tags: []
 
 # MERaLiON
 
-MERaLiON-AudioLLM is a Speech-Text Large Language Model tailored for Singapore’s multilingual and multicultural landscape. Integrating a localised Whisper-Large-V3 and SEA-LIONv3, MERaLiON-AudioLLM is finetuned on **260,000 hours of speech and audio data**, **8 various tasks**, to address the diverse linguistic nuances
-of Singapore local accents and dialects.
+MERaLiON-AudioLLM is a Speech-Text Large Language Model tailored for Singapore’s multilingual and multicultural landscape. Integrating a localised Whisper-large-v2 speech encoder and SEA-LIONv3 text decoder, MERaLiON-AudioLLM is finetuned on **260,000 hours of speech and audio data**, **8 various tasks**, to address the diverse linguistic nuances
+of Singapore's local accents and dialects.
 
 MERaLiON stands for <i>Multimodal Empathetic Reasoning and Learning in One Network</i>.
 
 - **Developed by:** I<sup>2</sup>R, A\*STAR
-- **Funded by [optional]:** Singapore NRF
+- **Funded by:** Singapore NRF
 - **Model type:** MultiModal LLM
-- **Language(s) (Speech):** English (general & Singapore)
+- **Language(s) (Speech):** English (Global & Singapore)
 - **Language(s) (NLP):** English, Chinese, Vietnamese, Indonesian, Thai, Filipino, Tamil, Malay, Khmer, Lao, Burmese, Javanese, Sundanese
-- **License:** MIT
+- **License:** [TBC]
 
 For more details, please refer to our [report]().
 
-## Model Details
+## Model Description
 
-### Model Description
+MERaLiON-AudioLLM is designed to take in an audio-text pair as
+input and generates a text output.
 
-<!-- Provide a longer summary of what this model is. -->
+The architecture comprises three key components: an audio encoder that transforms speech or audio inputs into sequences of vector representations, a text decoder that interprets and responds to natural language instructions, and an adaptor module that compresses the encoder representations while aligning the encoder’s hidden dimension with the text decoder’s embedding size.
 
+Specifically, we fine-tuned the MERaLiON-Whisper encoder from Whisper-large-v2 for the audio encoder and used SEA-LION V3, a localised LLM developed by our partner AI Singapore as the text decoder.
 
+## Capabilities
+
+MERaLiON-AudioLLM is trained to address 8 tasks, including Automatic Speech Recognition (ASR), Speech Translation (ST), Spoken Question Answering (SQA), Spoken Dialogue Summarization (SDS), Speech Instruction (SI), Paralinguistics (PARA), Audio Captioning (AC), and Audio Scene Question Answering (ASQA). 
+
+[More information about the 8 tasks and evaluation results]
 
 ## 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. -->
+Here we provide a code snippet illustrating the process of loading both the processor and model, alongside detailed instructions on executing the MERaLiON-AudioLLM model for content generation.
 
-### Direct Use
+**NOTE** This model has not been trained to use a system prompt or to use tool calling.
 
-<!-- This section is for the model use without fine-tuning or plugging into a larger ecosystem/app. -->
+### Inference
 
 ```python
-
 from datasets import load_dataset
 from transformers import AutoModelForSpeechSeq2Seq, AutoProcessor
 
@@ -52,13 +58,12 @@ model = AutoModelForSpeechSeq2Seq.from_pretrained(
     trust_remote_code=True,
 )
 
-prompt = "Can you please turn this audio into text format?"
+prompt = "Given the following audio context: <SpeechHere>\n\nText instruction: {query}"
+query = "Can you please turn this audio into text format?"
 conversation = [
-    {
-        "role": "user",
-        "content": f"Given the following audio context: <SpeechHere>\n\nText instruction: {prompt}"
-    }
+    {"role": "user", "content": prompt.format(query=query)}
 ]
+
 chat_prompt = processor.tokenizer.apply_chat_template(
     conversation=conversation,
     tokenize=False,
@@ -67,141 +72,75 @@ chat_prompt = processor.tokenizer.apply_chat_template(
 
 libri_data = load_dataset("distil-whisper/librispeech_long", "clean", split="validation")
 audio_array = libri_data[0]["audio"]["array"]
-
-inputs = processor(text=chat_prompt, audios=audio_array, time_duration_limit=20)
+inputs = processor(text=chat_prompt, audios=audio_array, time_duration_limit=30)
 
 outputs = model.generate(**inputs, max_new_tokens=128)
-
-print(processor.decode(outputs[0, inputs['input_ids'].size(1):], skip_special_tokens=True))
-
+generated_ids = outputs[:, inputs['input_ids'].size(1):]
+response = processor.batch_decode(generated_ids, skip_special_tokens=True)[0]
 ```
 
-### Downstream Use [optional]
+### Batch Inference
 
-<!-- This section is for the model use when fine-tuned for a task, or when plugged into a larger ecosystem/app -->
+MERaLiON-AudioLLM also supports batch inference.
 
-[More Information Needed]
-
-### Out-of-Scope Use
-
-<!-- This section addresses misuse, malicious use, and uses that the model will not work well for. -->
+```python
+from datasets import load_dataset
+from transformers import AutoModelForSpeechSeq2Seq, AutoProcessor
 
-[More Information Needed]
+repo_id = "MERaLiON/AudioLLM"
 
-## Bias, Risks, and Limitations
+processor = AutoProcessor.from_pretrained(
+    repo_id, 
+    trust_remote_code=True,
+    )
+model = AutoModelForSpeechSeq2Seq.from_pretrained(
+    repo_id,
+    use_safetensors=True,
+    trust_remote_code=True,
+)
 
-<!-- This section is meant to convey both technical and sociotechnical limitations. -->
+prompt = "Given the following audio context: <SpeechHere>\n\nText instruction: {query}"
+transcribe_query = "Can you please turn this speech into text format?"
+translate_query = "Can you please translate this speech into written Chinese?"
 
-[More Information Needed]
+conversation = [
+    [{"role": "user", "content": prompt.format(query=transcribe_query)}],
+    [{"role": "user", "content": prompt.format(query=translate_query)}],
+]
 
-### Recommendations
+chat_prompt = processor.tokenizer.apply_chat_template(
+    conversation=conversation,
+    tokenize=False,
+    add_generation_prompt=True
+)
 
-<!-- This section is meant to convey recommendations with respect to the bias, risk, and technical limitations. -->
+libri_data = load_dataset("distil-whisper/librispeech_long", "clean", split="validation")
+audio_array = [libri_data[0]["audio"]["array"]]*2
+inputs = processor(text=chat_prompt, audios=audio_array, time_duration_limit=30)
 
-Users (both direct and downstream) should be made aware of the risks, biases and limitations of the model. More information needed for further recommendations.
+outputs = model.generate(**inputs, max_new_tokens=128)
+generated_ids = outputs[:, inputs['input_ids'].size(1):]
+response = processor.batch_decode(generated_ids, skip_special_tokens=True)
+```
 
-## How to Get Started with the Model
+## Bias, Risks, and Limitations
 
-Use the code below to get started with the model.
+The current MERaLiON-AudioLLM has not been aligned for safety. Developers and users should perform their own safety fine-tuning and related security measures. In no event shall the authors be held liable for any claim, damages, or other liability arising from the use of the released weights and codes.
 
-[More Information Needed]
 
-## Training Details
+## Technical Specifications 
 
 ### Training Data
 
-<!-- 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. -->
-
-[More Information Needed]
-
-### Training Procedure
-
-<!-- This relates heavily to the Technical Specifications. Content here should link to that section when it is relevant to the training procedure. -->
-
-#### Preprocessing [optional]
-
-[More Information Needed]
-
-
-#### Training Hyperparameters
-
-- **Training regime:** [More Information Needed] <!--fp32, fp16 mixed precision, bf16 mixed precision, bf16 non-mixed precision, fp16 non-mixed precision, fp8 mixed precision -->
-
-#### Speeds, Sizes, Times [optional]
-
-<!-- This section provides information about throughput, start/end time, checkpoint size if relevant, etc. -->
-
-[More Information Needed]
-
-## Evaluation
-
-<!-- This section describes the evaluation protocols and provides the results. -->
-
-### Testing Data, Factors & Metrics
-
-#### Testing Data
-
-<!-- This should link to a Dataset Card if possible. -->
-
-[More Information Needed]
-
-#### Factors
-
-<!-- These are the things the evaluation is disaggregating by, e.g., subpopulations or domains. -->
-
-[More Information Needed]
-
-#### Metrics
-
-<!-- These are the evaluation metrics being used, ideally with a description of why. -->
-
-[More Information Needed]
-
-### Results
-
-[More Information Needed]
-
-#### Summary
-
-
-
-## Model Examination [optional]
-
-<!-- Relevant interpretability work for the model goes here -->
-
-[More Information Needed]
-
-## Environmental Impact
+MERaLiON-AudioLLM is trained on a diverse collection of publicly available datasets, alongside synthesised and augmented samples carefully curated by the team and native speakers, totaling 260,000 hours of audio.
 
-<!-- Total emissions (in grams of CO2eq) and additional considerations, such as electricity usage, go here. Edit the suggested text below accordingly -->
+### Compute and Infrastructure
 
-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).
+MERaLiON-AudioLLM is trained on the ASPIRE 2A+ Supercomputer Cluster, provided by the National Supercomputing Centre (NSCC). ASPIRE 2A+ cluster provides multiple H100 nodes, with each compute node equipped with 8 Nvidia H100 GPUs, 2 TB of RAM, and 30 TB of locally attached NVMe storage. These nodes are interconnected via a rail-optimised, full fat-tree topology, utilising 400 Gb/s NDR InfiniBand cables. Additionally, the cluster incorporates a 2.5 PB SSD-based Lustre file system, linked to the H100 nodes through high-speed InfiniBand connections. 
 
-- **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]
+With a global batch size of 640, we train the current release of MERaLiON-AudioLLM for around 200k steps, which took 2 days to complete using 16 nodes, 128 H100 GPUs.
 
-## Technical Specifications [optional]
-
-### Model Architecture and Objective
-
-[More Information Needed]
-
-### Compute Infrastructure
-
-[More Information Needed]
-
-#### Hardware
-
-[More Information Needed]
-
-#### Software
-
-[More Information Needed]
-
-## Citation [optional]
+## Citation
 
 <!-- If there is a paper or blog post introducing the model, the APA and Bibtex information for that should go in this section. -->
 
@@ -211,22 +150,4 @@ Carbon emissions can be estimated using the [Machine Learning Impact calculator]
 
 **APA:**
 
-[More Information Needed]
-
-## Glossary [optional]
-
-<!-- If relevant, include terms and calculations in this section that can help readers understand the model or model card. -->
-
-[More Information Needed]
-
-## More Information [optional]
-
-[More Information Needed]
-
-## Model Card Authors [optional]
-
-[More Information Needed]
-
-## Model Card Contact
-
 [More Information Needed]