updated notebook for newer setup structure, added formatting of getting it/s and lass loss rate (have not tested loss rate yet)

This commit is contained in:
mrq 2023-02-20 22:56:39 +00:00
parent bacac6daea
commit 1fd88afcca
2 changed files with 41 additions and 18 deletions

View File

@ -41,17 +41,20 @@
"!git clone https://git.ecker.tech/mrq/ai-voice-cloning/\n", "!git clone https://git.ecker.tech/mrq/ai-voice-cloning/\n",
"%cd ai-voice-cloning\n", "%cd ai-voice-cloning\n",
"\n", "\n",
"!git submodule init\n",
"!git submodule update\n",
"\n",
"# TODO: fix venvs working for subprocess.Popen calling a bash script\n", "# TODO: fix venvs working for subprocess.Popen calling a bash script\n",
"#!apt install python3.8-venv\n", "#!apt install python3.8-venv\n",
"#!python -m venv venv\n", "#!python -m venv venv\n",
"#!source ./venv/bin/activate\n", "#!source ./venv/bin/activate\n",
"\n", "\n",
"!git clone https://git.ecker.tech/mrq/DL-Art-School dlas\n",
"!python -m pip install --upgrade pip\n", "!python -m pip install --upgrade pip\n",
"!pip install torch torchvision torchaudio --extra-index-url https://download.pytorch.org/whl/cu116\n", "!pip install torch torchvision torchaudio --extra-index-url https://download.pytorch.org/whl/cu116\n",
"!./setup-tortoise.sh\n", "!python -m pip install -r ./dlas/requirements.txt\n",
"!./setup-training.sh\n", "!python -m pip install -r ./tortoise-tts/requirements.txt\n",
"!python -m pip install -r ./requirements.txt" "!python -m pip install -r ./requirements.txt\n",
"!python -m pip install -e ./tortoise-tts/"
] ]
}, },
{ {
@ -67,15 +70,8 @@
"cell_type":"code", "cell_type":"code",
"source":[ "source":[
"# for my debugging purposes\n", "# for my debugging purposes\n",
"%cd /content/ai-voice-cloning/dlas\n", "%cd /content/ai-voice-cloning/\n",
"!git reset --hard HEAD\n", "!./update.sh\n",
"!git pull\n",
"%cd ../tortoise-tts/\n",
"!git reset --hard HEAD\n",
"!git pull\n",
"!cd ..\n",
"!git reset --hard HEAD\n",
"!git pull\n",
"# exit()" "# exit()"
], ],
"metadata":{ "metadata":{

View File

@ -417,9 +417,16 @@ def run_training(config_path, verbose=False, buffer_size=8, progress=gr.Progress
yield " ".join(cmd) yield " ".join(cmd)
info = {}
buffer = [] buffer = []
infos = [] infos = []
yields = True yields = True
status = ""
it_rate = ""
it_time_start = 0
it_time_end = 0
for line in iter(training_process.stdout.readline, ""): for line in iter(training_process.stdout.readline, ""):
buffer.append(f'{line}') buffer.append(f'{line}')
@ -430,13 +437,34 @@ def run_training(config_path, verbose=False, buffer_size=8, progress=gr.Progress
elif progress is not None: elif progress is not None:
if line.find(' 0%|') == 0: if line.find(' 0%|') == 0:
open_state = True open_state = True
it_time_start = time.time()
elif line.find('100%|') == 0 and open_state: elif line.find('100%|') == 0 and open_state:
it_time_end = time.time()
open_state = False open_state = False
it = it + 1 it = it + 1
progress(it / float(its), f'[{it}/{its}] Training...')
elif line.find('INFO: [epoch:') >= 0: it_time_delta = it_time_end-it_time_start
infos.append(f'{line}') it_rate = f'[{"{:.3f}".format(it_time_delta)}s/it]' if it_time_delta >= 1 and it_time_delta != 0 else f'[{"{:.3f}".format(1/it_time_delta)}it/s]' # I doubt anyone will have it/s rates, but its here
elif line.find('Saving models and training states') >= 0:
progress(it / float(its), f'[{it}/{its}] {it_rate} Training... {status}')
# try because I haven't tested this yet
try:
if line.find('INFO: [epoch:') >= 0:
# easily rip out our stats...
match = re.findall(r'\b([a-z_0-9]+?)\b: ([0-9]\.[0-9]+?e[+-]\d+)\b', line)
if match and len(match) > 0:
for k, v in match:
info[k] = float(v)
# ...and returns our loss rate
# it would be nice for losses to be shown at every step
if 'loss_gpt_total' in info:
status = f"Total loss at step {int(info['step'])}: {info['loss_gpt_total']}"
except Exception as e:
pass
if line.find('Saving models and training states') >= 0:
checkpoint = checkpoint + 1 checkpoint = checkpoint + 1
progress(checkpoint / float(checkpoints), f'[{checkpoint}/{checkpoints}] Saving checkpoint...') progress(checkpoint / float(checkpoints), f'[{checkpoint}/{checkpoints}] Saving checkpoint...')
@ -459,7 +487,6 @@ def stop_training():
if training_process is None: if training_process is None:
return "No training in progress" return "No training in progress"
training_process.kill() training_process.kill()
training_process = None
return "Training cancelled" return "Training cancelled"
def prepare_dataset( files, outdir, language=None, progress=None ): def prepare_dataset( files, outdir, language=None, progress=None ):