textrnn-pytorch

nvidia-docker

nvidia-docker run -v /home/clay/workspace:/workspace --shm-size=1g --ulimit memlock=-1 --ulimit stack=67108864 -it --rm nvcr.io/nvidia/tensorflow:19.05-py3

training

python train.py --mode train --file ./training/metal_song_titles/source/The-Collected-Works-of-HP-Lovecraft_djvu_poems_clean.txt --session metal03 --number 4000

prediction

python train.py --mode predict --session metal03

python train.py --mode predict --file ./training/metal_song_titles/source/The-Collected-Works-of-HP-Lovecraft_djvu_poems_clean.txt --session metal04 --initial "I am very busy"

__Python VERSION: 3.7.7 (default, May 7 2020, 21:25:33) [GCC 7.3.0] __pyTorch VERSION: 1.3.1 __CUDA AVILABLE: True __CUDNN VERSION: 7605 __Number CUDA Devices: 1 __Device: cuda GeForce RTX 2080 Ti Active CUDA Device: GPU 0 Available devices 1 Current cuda device 0 Vocabulary size 5778 High This Hybrid Valiant Tho' Deeds Flutter

running the title app

FLASK_APP=app.py APP_CONFIG=textrnn.cfg flask run --host=0.0.0.0 --port=5001

curl to get text from the the tile app

curl http://localhost:5001/title?session_id=metal04

{"title":"Glimpse No Wind Swear"}

training using the NVIDIA docker container

docker run --gpus all --shm-size=1g --ulimit memlock=-1
--ulimit stack=67108864 -it --rm
-v $(pwd)/workspace:/workspace nvcr.io/nvidia/pytorch:20.06-py3
python train.py --mode train
--file /workspace/training/hplc/metal_gothic_poetry.txt
--session metal03 --number 4000

building the docker container

docker build . -t textrnn:latest

train

docker run --gpus all --shm-size=1g --ulimit memlock=-1 --ulimit stack=67108864 -it --rm -v $(pwd)/workspace:/workspace textrnn:latest python src/train.py --mode train --file /workspace/training/hplc/metal_gothic_poetry.txt --session metal05 --number 1000

docker run --gpus all --shm-size=1g --ulimit memlock=-1 --ulimit stack=67108864 -it --rm -v $(pwd)/workspace:/workspace textrnn:latest python src/train.py --mode train --file /workspace/training/hplc/metal_gothic_poetry.txt --session reflect_01 --number 40000

predict

docker run --gpus all --shm-size=1g --ulimit memlock=-1 --ulimit stack=67108864 -it --rm -v $(pwd)/workspace:/workspace textrnn:latest python src/train.py --mode predict --session metal05

and

docker run --gpus all --shm-size=1g --ulimit memlock=-1 --ulimit stack=67108864 -it --rm -v $(pwd)/workspace:/workspace textrnn:latest python src/predict.py --session metal05 --predict 500 --lines 10

and

docker run --gpus all --shm-size=1g --ulimit memlock=-1 --ulimit stack=67108864 -it --rm  -v $(pwd)/workspace:/workspace textrnn:latest python src/predict.py --session metal05 --predict 500 --lines 10

docker run --gpus all --shm-size=1g --ulimit memlock=-1 --ulimit stack=67108864 -it --rm -v $(pwd)/workspace:/workspace textrnn:latest python src/predict.py --session metal05 --predic 500 --lines 10 --initial "disquieting earth" -o /workspace/reflect/mar_3_2021

docker run --gpus all --shm-size=1g --ulimit memlock=-1 --ulimit stack=67108864 -it --rm -v $(pwd)/workspace:/workspace textrnn:latest python src/generate.py --session metal05 --predic 800 --lines 20 --initial "disquieting earth"

python src/generate.py --session reflect_02 --predic 800 --initial "disquieting earth"

python src/predict.py --session reflect_02 --predic 500 --lines 10 --initial "disquieting earth"

run the flask app

cli

python -m flask run --host=0.0.0.0 --port=8002

as a docker instance

docker run -p 8002:8002 --gpus all --shm-size=1g --ulimit memlock=-1 --ulimit stack=67108864 -it --rm  -v $(pwd)/workspace:/workspace textrnn:latest

claytantor/textrnn-pytorch