Sort dropdowns numerically

This commit is contained in:
oobabooga 2023-05-05 23:14:56 -03:00
parent ee3c8a893e
commit 56f6b7052a
3 changed files with 95 additions and 83 deletions

View file

@ -14,9 +14,9 @@ from datasets import Dataset, load_dataset
from peft import (LoraConfig, get_peft_model, prepare_model_for_int8_training,
set_peft_model_state_dict)
from modules import shared, ui
from modules import shared, ui, utils
from modules.evaluate import calculate_perplexity, generate_markdown_table, save_past_evaluations
from server import get_available_loras, get_available_models
# This mapping is from a very recent commit, not yet released.
# If not available, default to a backup map for some common model types.
@ -41,10 +41,6 @@ WANT_INTERRUPT = False
PARAMETERS = ["lora_name", "always_override", "save_steps", "micro_batch_size", "batch_size", "epochs", "learning_rate", "lr_scheduler_type", "lora_rank", "lora_alpha", "lora_dropout", "cutoff_len", "dataset", "eval_dataset", "format", "eval_steps", "raw_text_file", "overlap_len", "newline_favor_len", "higher_rank_limit", "warmup_steps", "optimizer"]
def get_datasets(path: str, ext: str):
return ['None'] + sorted(set([k.stem for k in Path(path).glob(f'*.{ext}') if k.stem != 'put-trainer-datasets-here']), key=str.lower)
def create_train_interface():
with gr.Tab('Train LoRA', elem_id='lora-train-tab'):
gr.Markdown("Confused? [[Click here for a guide]](https://github.com/oobabooga/text-generation-webui/blob/main/docs/Training-LoRAs.md)")
@ -55,8 +51,8 @@ def create_train_interface():
save_steps = gr.Number(label='Save every n steps', value=0, info='If above 0, a checkpoint of the LoRA will be saved every time this many steps pass.')
with gr.Row():
copy_from = gr.Dropdown(label='Copy parameters from', value='None', choices=get_available_loras())
ui.create_refresh_button(copy_from, lambda: None, lambda: {'choices': get_available_loras()}, 'refresh-button')
copy_from = gr.Dropdown(label='Copy parameters from', value='None', choices=utils.get_available_loras())
ui.create_refresh_button(copy_from, lambda: None, lambda: {'choices': utils.get_available_loras()}, 'refresh-button')
with gr.Row():
# TODO: Implement multi-device support.
@ -76,19 +72,19 @@ def create_train_interface():
with gr.Tab(label='Formatted Dataset'):
with gr.Row():
dataset = gr.Dropdown(choices=get_datasets('training/datasets', 'json'), value='None', label='Dataset', info='The dataset file to use for training.')
ui.create_refresh_button(dataset, lambda: None, lambda: {'choices': get_datasets('training/datasets', 'json')}, 'refresh-button')
eval_dataset = gr.Dropdown(choices=get_datasets('training/datasets', 'json'), value='None', label='Evaluation Dataset', info='The (optional) dataset file used to evaluate the model after training.')
ui.create_refresh_button(eval_dataset, lambda: None, lambda: {'choices': get_datasets('training/datasets', 'json')}, 'refresh-button')
format = gr.Dropdown(choices=get_datasets('training/formats', 'json'), value='None', label='Data Format', info='The format file used to decide how to format the dataset input.')
ui.create_refresh_button(format, lambda: None, lambda: {'choices': get_datasets('training/formats', 'json')}, 'refresh-button')
dataset = gr.Dropdown(choices=utils.get_datasets('training/datasets', 'json'), value='None', label='Dataset', info='The dataset file to use for training.')
ui.create_refresh_button(dataset, lambda: None, lambda: {'choices': utils.get_datasets('training/datasets', 'json')}, 'refresh-button')
eval_dataset = gr.Dropdown(choices=utils.get_datasets('training/datasets', 'json'), value='None', label='Evaluation Dataset', info='The (optional) dataset file used to evaluate the model after training.')
ui.create_refresh_button(eval_dataset, lambda: None, lambda: {'choices': utils.get_datasets('training/datasets', 'json')}, 'refresh-button')
format = gr.Dropdown(choices=utils.get_datasets('training/formats', 'json'), value='None', label='Data Format', info='The format file used to decide how to format the dataset input.')
ui.create_refresh_button(format, lambda: None, lambda: {'choices': utils.get_datasets('training/formats', 'json')}, 'refresh-button')
eval_steps = gr.Number(label='Evaluate every n steps', value=100, info='If an evaluation dataset is given, test it every time this many steps pass.')
with gr.Tab(label="Raw text file"):
with gr.Row():
raw_text_file = gr.Dropdown(choices=get_datasets('training/datasets', 'txt'), value='None', label='Text file', info='The raw text file to use for training.')
ui.create_refresh_button(raw_text_file, lambda: None, lambda: {'choices': get_datasets('training/datasets', 'txt')}, 'refresh-button')
raw_text_file = gr.Dropdown(choices=utils.get_datasets('training/datasets', 'txt'), value='None', label='Text file', info='The raw text file to use for training.')
ui.create_refresh_button(raw_text_file, lambda: None, lambda: {'choices': utils.get_datasets('training/datasets', 'txt')}, 'refresh-button')
with gr.Row():
overlap_len = gr.Slider(label='Overlap Length', minimum=0, maximum=512, value=128, step=16, info='Overlap length - ie how many tokens from the prior chunk of text to include into the next chunk. (The chunks themselves will be of a size determined by Cutoff Length below). Setting overlap to exactly half the cutoff length may be ideal.')
@ -111,8 +107,8 @@ def create_train_interface():
with gr.Tab('Perplexity evaluation', elem_id='evaluate-tab'):
with gr.Row():
with gr.Column():
models = gr.Dropdown(get_available_models(), label='Models', multiselect=True)
evaluate_text_file = gr.Dropdown(choices=['wikitext', 'ptb', 'ptb_new'] + get_datasets('training/datasets', 'txt')[1:], value='wikitext', label='Input dataset', info='The raw text file on which the model will be evaluated. The first options are automatically downloaded: wikitext, ptb, and ptb_new. The next options are your local text files under training/datasets.')
models = gr.Dropdown(utils.get_available_models(), label='Models', multiselect=True)
evaluate_text_file = gr.Dropdown(choices=['wikitext', 'ptb', 'ptb_new'] + utils.get_datasets('training/datasets', 'txt')[1:], value='wikitext', label='Input dataset', info='The raw text file on which the model will be evaluated. The first options are automatically downloaded: wikitext, ptb, and ptb_new. The next options are your local text files under training/datasets.')
with gr.Row():
stride_length = gr.Slider(label='Stride', minimum=1, maximum=2048, value=512, step=1, info='Used to make the evaluation faster at the cost of accuracy. 1 = slowest but most accurate. 512 is a common value.')
max_length = gr.Slider(label='max_length', minimum=0, maximum=8096, value=0, step=1, info='The context for each evaluation. If set to 0, the maximum context length for the model will be used.')

61
modules/utils.py Normal file
View file

@ -0,0 +1,61 @@
import os
import re
from pathlib import Path
from modules import shared
def atoi(text):
return int(text) if text.isdigit() else text.lower()
def natural_keys(text):
return [atoi(c) for c in re.split(r'(\d+)', text)]
def get_available_models():
if shared.args.flexgen:
return sorted([re.sub('-np$', '', item.name) for item in list(Path(f'{shared.args.model_dir}/').glob('*')) if item.name.endswith('-np')], key=natural_keys)
else:
return sorted([re.sub('.pth$', '', item.name) for item in list(Path(f'{shared.args.model_dir}/').glob('*')) if not item.name.endswith(('.txt', '-np', '.pt', '.json', '.yaml'))], key=natural_keys)
def get_available_presets():
return sorted(set((k.stem for k in Path('presets').glob('*.txt'))), key=natural_keys)
def get_available_prompts():
prompts = []
prompts += sorted(set((k.stem for k in Path('prompts').glob('[0-9]*.txt'))), key=natural_keys, reverse=True)
prompts += sorted(set((k.stem for k in Path('prompts').glob('*.txt'))), key=natural_keys)
prompts += ['None']
return prompts
def get_available_characters():
paths = (x for x in Path('characters').iterdir() if x.suffix in ('.json', '.yaml', '.yml'))
return ['None'] + sorted(set((k.stem for k in paths if k.stem != "instruction-following")), key=natural_keys)
def get_available_instruction_templates():
path = "characters/instruction-following"
paths = []
if os.path.exists(path):
paths = (x for x in Path(path).iterdir() if x.suffix in ('.json', '.yaml', '.yml'))
return ['None'] + sorted(set((k.stem for k in paths)), key=natural_keys)
def get_available_extensions():
return sorted(set(map(lambda x: x.parts[1], Path('extensions').glob('*/script.py'))), key=natural_keys)
def get_available_softprompts():
return ['None'] + sorted(set((k.stem for k in Path('softprompts').glob('*.zip'))), key=natural_keys)
def get_available_loras():
return sorted([item.name for item in list(Path(shared.args.lora_dir).glob('*')) if not item.name.endswith(('.txt', '-np', '.pt', '.json'))], key=natural_keys)
def get_datasets(path: str, ext: str):
return ['None'] + sorted(set([k.stem for k in Path(path).glob(f'*.{ext}') if k.stem != 'put-trainer-datasets-here']), key=natural_keys)