m-a-p/MuPT-v0-8192-550M

SMuPT: Symbolic Music Generative Pre-trained Transformer

SMuPT is a series of pre-trained models for symbolic music generation. It was trained on a large-scale dataset of symbolic music, including millions of monophonic and polyphonic pieces from different genres and styles. The models are trained with the LLama2 architecture, and can be further used for downstream music generation tasks such as melody generation, accompaniment generation, and multi-track music generation.

• 09/01/2024: a series of pre-trained SMuPT models are released, with parameters ranging from 110M to 1.3B.

Model architecture

The details of model architecture of SMuPT-v0 are listed below:

Name	Parameters	Training Data(Music Pieces)	Seq Length	Hidden Size	Layers	Heads
SMuPT-v0-8192-110M	110M	7M x 5.8 epochs	8192	768	12	12
SMuPT-v0-8192-345M	345M	7M x 4 epochs	8192	1024	24	16
SMuPT-v0-8192-770M	770M	7M x 3 epochs	8192	1280	36	20
SMuPT-v0-8192-1.3B	1.3B	7M x 2.2 epochs	8192	1536	48	24

Model Usage

There are several ways to use our pre-trained SMuPT models, we now the usage based on Megatron-LM. Huggingface format will be supported soon.

Before starting, make sure you have setup the relevant environment and codebase.

# pull Megatron-LM codebase
mkdir -p /path/to/workspace && cd /path/to/workspace
git clone https://github.com/NVIDIA/Megatron-LM.git

# download the pre-trained SMuPT models checkpoint and vocab files from Huggingface page
mkdir -p /models/SMuPT_v0_8192_1.3B && cd /models/SMuPT_v0_8192_1.3B
wget -O model_optim_rng.pt https://huggingface.co/m-a-p/SMuPT_v0_8192_1.3B/resolve/main/model_optim_rng.pt?download=true
wget -O newline.vocab https://huggingface.co/m-a-p/SMuPT_v0_8192_1.3B/resolve/main/newline.vocab?download=true
wget -O newline.txt https://huggingface.co/m-a-p/SMuPT_v0_8192_1.3B/resolve/main/newline.txt?download=true

We recommend using the latest version of NGC's PyTorch container for SMuPT inference. See more details in Megatron-LM

# pull the latest NGC's PyTorch container, mount the workspace directory and enter the container
docker run --gpus all -it --name megatron --shm-size=16g -v $PWD:/workspace -p 5000:5000 nvcr.io/nvidia/pytorch:23.11-py3 /bin/bash

Once you enter the container, you can start a REST server for inference.

Click to expand the example script

#!/bin/bash
# This example will start serving the 1.3B model.
export CUDA_DEVICE_MAX_CONNECTIONS=1

DISTRIBUTED_ARGS="--nproc_per_node 1 \
                --nnodes 1 \
                --node_rank 0 \
                --master_addr localhost \
                --master_port 6000"

CHECKPOINT=/path/to/model/checkpoint/folder
VOCAB_FILE=/path/to/vocab/file
MERGE_FILE=/path/to/merge/file

MODEL_SIZE="1.3B"
if   [[ ${MODEL_SIZE} == "110M" ]];   then HIDDEN_SIZE=768;  NUM_HEAD=12; NUM_QUERY_GROUP=12; NUM_LAYERS=12; FFN_HIDDEN_SIZE=3072; NORM_EPS=1e-5;
elif [[ ${MODEL_SIZE} == "345M" ]];   then HIDDEN_SIZE=1024;  NUM_HEAD=16; NUM_QUERY_GROUP=16; NUM_LAYERS=24; FFN_HIDDEN_SIZE=4096; NORM_EPS=1e-5;
elif [[ ${MODEL_SIZE} == "770M" ]];   then HIDDEN_SIZE=1280;  NUM_HEAD=20; NUM_QUERY_GROUP=20; NUM_LAYERS=36; FFN_HIDDEN_SIZE=5120; NORM_EPS=1e-5;
elif [[ ${MODEL_SIZE} == "1.3B" ]];   then HIDDEN_SIZE=1536;  NUM_HEAD=24; NUM_QUERY_GROUP=24; NUM_LAYERS=48; FFN_HIDDEN_SIZE=6144; NORM_EPS=1e-5;
else echo "invalid MODEL_SIZE: ${MODEL_SIZE}"; exit 1
fi
MAX_SEQ_LEN=8192
MAX_POSITION_EMBEDDINGS=8192

pip install flask-restful

torchrun $DISTRIBUTED_ARGS tools/run_text_generation_server.py   \
    --tensor-model-parallel-size 1  \
    --pipeline-model-parallel-size 1  \
    --num-layers ${NUM_LAYERS}  \
    --hidden-size ${HIDDEN_SIZE}  \
    --ffn-hidden-size ${FFN_HIDDEN_SIZE} \
    --load ${CHECKPOINT}  \
    --group-query-attention \
    --num-query-groups ${NUM_QUERY_GROUP} \
    --position-embedding-type rope \
    --num-attention-heads ${NUM_HEAD}  \
    --max-position-embeddings ${MAX_POSITION_EMBEDDINGS}  \
    --tokenizer-type GPT2BPETokenizer  \
    --normalization RMSNorm \
    --norm-epsilon ${NORM_EPS} \
    --make-vocab-size-divisible-by 1 \
    --swiglu \
    --use-flash-attn \
    --bf16  \
    --micro-batch-size 1  \
    --disable-bias-linear \
    --no-bias-gelu-fusion \
    --untie-embeddings-and-output-weights \
    --seq-length ${MAX_SEQ_LEN}  \
    --vocab-file $VOCAB_FILE  \
    --merge-file $MERGE_FILE  \
    --attention-dropout 0.0 \
    --hidden-dropout 0.0 \
    --weight-decay 1e-1 \
    --clip-grad 1.0 \
    --adam-beta1 0.9 \
    --adam-beta2 0.95 \
    --adam-eps 1e-8 \
    --seed 42

Use CURL to query the server directly, note that the newline token \n is represented by <n> in the vocabulary, so we need to replace the newline token with <n> in both the prompt and the generated tokens.

curl 'http://localhost:6000/api' -X 'PUT' -H 'Content-Type: application/json; charset=UTF-8'  -d '{"prompts":["X:1<n>T:Music21 Fragment<n>T:Music21 Fragment<n>T:Music21<n>C:Music21<n>%%score 1 2 3 4<n>L:1/4<n>M:4/4<n>K:C<n>V:1 treble nm=\"Piano\" snm=\"Pno\"<n>%%MIDI program 0<n>%%MIDI control 7 100<n>%%MIDI control 10 64<n>V:2 treble nm=\"Piano\" snm=\"Pno\"<n>%%MIDI channel 3<n>%%MIDI program 0<n>%%MIDI control 7 100<n>%%MIDI control 10 64<n>V:3 bass nm=\"Piano\" snm=\"Pno\"<n>%%MIDI channel 4<n>%%MIDI program 0<n>%%MIDI control 7 100<n>%%MIDI control 10 64<n>V:4 bass nm=\"Piano\" snm=\"Pno\"<n>%%MIDI channel 5<n>%%MIDI program 0<n>%%MIDI control 7 100<n>%%MIDI control 10 64<n>V:1<n> z3 c | B A G F/E/ | !fermata!E ^F ^G A | B e d c | !fermata!B2 G A | B c d e | %6<n> !fermata!c2"], "tokens_to_generate":4096}'

Processed Output:

X:1
T:Music21 Fragment
T:Music21 Fragment
T:Music21
C:Music21
%%score 1 2 3 4
L:1/4
M:4/4
K:C
V:1 treble nm="Piano" snm="Pno"
%%MIDI program 0
%%MIDI control 7 100
%%MIDI control 10 64
V:2 treble nm="Piano" snm="Pno"
%%MIDI channel 3
%%MIDI program 0
%%MIDI control 7 100
%%MIDI control 10 64
V:3 bass nm="Piano" snm="Pno"
%%MIDI channel 4
%%MIDI program 0
%%MIDI control 7 100
%%MIDI control 10 64
V:4 bass nm="Piano" snm="Pno"
%%MIDI channel 5
%%MIDI program 0
%%MIDI control 7 100
%%MIDI control 10 64
V:1
 z3 c | B A G F/E/ | !fermata!E ^F ^G A | B e d c | !fermata!B2 G A | B c d e | %6
 !fermata!c2 e | a g/f/ e d | !fermata!e2- e e | a e a g/f/ | e2 !fermata!d g | d d g e |1
 e2 !fermata!e G :|3 e2 !fermata!e2- || e z z2 | z4 |]
V:2
 z3 G | G E C B, | !fermata!C D E E | G G ^F A | !fermata!^G2 E E | G G G G | sG !fermata!B2 G |
 c B B A | !fermata!^G2- G G | A B c d | c2 !fermata!B _B | A A _A A |1 G2 A !fermata!G :|3
 G2 !fermata!A2- || A z z2 | z4 |]
V:3
 z3 C,/D,/ | E, E, E, ^G,, | !fermata!A,, B,, B,, C,/D,/ | E, C, D, ^F, | !fermata!E,2 C, C,/D,/ |
 E, C, B,, C, | !fermata!G, !fermata!G,,2 C,/D,/ | E, E, F, D, | !fermata!E,2- E, D, |
 C, B,, A,, B,, | C,3/2 D,/ !fermata!G,, E, | F, F, _E, B,, |1 _B,, G,, ^F,, !fermata!G,, :|3
 _B,, G,, !fermata!^F,,2- || F,, z z2 | z4 |]
V:4
 z3 C,/B,,/ | A,, E,, A,, E,, | !fermata!A,, G,, E,, A,, | G,, C, B,, D,, |
 !fermata!E,,2 E,,/F,,/ A,, | G,, C, G,, C, | !fermata!B,, !fermata!G,,2 C,/B,,/ | A,, E,, F,, D,, |
 !fermata!E,,2- E,,/^F,,/ ^G,, | A,, ^G,, A,, B,, | C,3/2 D,/ !fermata!G,, E,, |
 F,,/G,,/ A,, B,, E,, |1 G,, C,, ^F,, !fermata!G,, :|3 G,, C,, !fermata!^F,,2- || F,, z z2 | z4 |]

Once you encode the generated tokens into audio, you will hear the following music.