Mirrors/text-generation-webui
1
0
Fork 0
mirror of https://github.com/oobabooga/text-generation-webui.git synced 2024-11-01 07:00:15 +01:00
Code Issues Packages Projects Releases Wiki Activity
6a61158adf
text-generation-webui/extensions/Training_PRO/script.py
FartyPants (FP HAM) 6a61158adf
Training PRO a month worth of updates (#4345)
2023-10-22 12:38:09 -03:00

1377 lines
71 KiB
Python
Raw Blame History

import os
os.environ["WANDB_MODE"] = "offline"
# os.environ["WANDB_DISABLED"] = "true"
import json
import math
import random
import shutil
import sys
import threading
import time
import traceback
from datetime import datetime
from pathlib import Path
import gradio as gr
import pandas as pd
import torch
import transformers
from functools import partial
from .custom_scheduler import FPSchedulerTrainer, FPNEFtuneTrainer
from .matplotgraph import create_graph
from .train_utils import get_available_loras_local, precise_cut, sliding_block_cut, download_file_from_url
from datasets import Dataset, load_dataset
from peft import (
LoraConfig,
get_peft_model,
prepare_model_for_kbit_training,
set_peft_model_state_dict
)
from peft.utils.other import \
TRANSFORMERS_MODELS_TO_LORA_TARGET_MODULES_MAPPING as model_to_lora_modules
from transformers.models.auto.modeling_auto import (
MODEL_FOR_CAUSAL_LM_MAPPING_NAMES
)
from modules import shared, utils
from modules.ui import create_refresh_button
from modules.evaluate import (
calculate_perplexity,
generate_markdown_table,
save_past_evaluations
)
from modules.logging_colors import logger
from modules.models import reload_model
from modules.utils import natural_keys
## just temporary to avoid warning
import inspect
from typing import Callable, Optional, Tuple, ContextManager
if hasattr(torch.utils.checkpoint, 'noop_context_fn'):
def my_checkpoint(
function,
*args,
use_reentrant: Optional[bool] = None,
context_fn: Callable[[], Tuple[ContextManager, ContextManager]] = torch.utils.checkpoint.noop_context_fn,
determinism_check: str = torch.utils.checkpoint._DEFAULT_DETERMINISM_MODE,
debug: bool = False,
**kwargs
):
if use_reentrant is None:
#print ("reentran = NONE")
use_reentrant = True
# Hack to mix *args with **kwargs in a python 2.7-compliant way
preserve = kwargs.pop("preserve_rng_state", True)
if kwargs and use_reentrant:
raise ValueError(
"Unexpected keyword arguments: " + ",".join(arg for arg in kwargs)
)
if use_reentrant:
if context_fn is not torch.utils.checkpoint.noop_context_fn or debug is not False:
raise ValueError(
"Passing `context_fn` or `debug` is only supported when "
"use_reentrant=False."
)
return torch.utils.checkpoint.CheckpointFunction.apply(function, preserve, *args)
else:
print ("reentran = FALSE")
gen = torch.utils.checkpoint._checkpoint_without_reentrant_generator(
function, preserve, context_fn, determinism_check, debug, *args, **kwargs
)
# Runs pre-forward logic
next(gen)
ret = function(*args, **kwargs)
# Runs post-forward logic
try:
next(gen)
except StopIteration:
return ret
params = {
"display_name": "Training PRO",
"is_tab": True
}
non_serialized_params = {
"debug_slicer": False,
"Lora_sortedByTime": False,
"stop_at_loss": 0,
"save_steps_under_loss": 0.0,
"save_checkpoint_now": False,
"training_loop": False,
"current_stability": 0,
"save_epochs": 0,
"checkpoint_offset": 0,
"epoch_offset":0,
}
MODEL_CLASSES = {v[1]: v[0] for v in MODEL_FOR_CAUSAL_LM_MAPPING_NAMES.items()}
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", "higher_rank_limit", "warmup_steps", "optimizer", "hard_cut_string", "train_only_after", "stop_at_loss", "add_eos_token", "min_chars", "report_to", "precize_slicing_overlap", "add_eos_token_type", "save_steps_under_loss", "add_bos_token", "training_projection","sliding_window","warmup_ratio","grad_accumulation","neft_noise_alpha"]
WANT_INTERRUPT = False
train_log = {}
train_template = {}
train_log_graph = []
train_choices = ["all","q-k-v-o","q-k-v","k-v-down","q-v"]
statistics = {
'loss': [],
'lr': [],
}
RED = "\033[91m"
YELLOW = "\033[93m"
GREEN = "\033[92m"
RESET = "\033[0m"
def ui():
with gr.Tab('Train LoRA', elem_id='lora-train-tab'):
tmp = gr.State('')
with gr.Row():
with gr.Column():
# YY.MM.DD
gr.Markdown("`Ver: 23.10.20` This is enhanced version of QLora Training. [Maintained by FP](https://github.com/FartyPants/Training_PRO/tree/main)")
with gr.Row():
with gr.Column(scale=5):
with gr.Row():
copy_from = gr.Dropdown(label='Copy parameters from', value='None', choices=get_available_loras_local(non_serialized_params['Lora_sortedByTime']), elem_classes=['slim-dropdown'])
create_refresh_button(copy_from, lambda: None, lambda: {'choices': get_available_loras_local(non_serialized_params['Lora_sortedByTime'])}, 'refresh-button')
with gr.Column():
sort_byTime = gr.Checkbox(label='Sort list by Date', value=False, info='Sorts Loras by date created.', elem_classes=['no-background'])
with gr.Row():
with gr.Column(scale=5):
lora_name = gr.Textbox(label='Name', info='The name of your new LoRA file')
with gr.Column():
always_override = gr.Checkbox(label='Override Existing Files', value=False, info='If the name is the same, checking will replace the existing file, and unchecking will load and continue from it (the rank must be the same).', elem_classes=['no-background'])
with gr.Row():
with gr.Column():
lora_rank = gr.Slider(label='LoRA Rank', value=32, minimum=0, maximum=1024, step=4, info='Also called dimension count. Higher values = larger file, more content control. Smaller values = smaller file, less control. Use 4 or 8 for style, 128 or 256 to teach, 1024+ for fine-detail on big data. More VRAM is needed for higher ranks.')
lora_alpha = gr.Slider(label='LoRA Alpha', value=64, minimum=0, maximum=2048, step=4, info='This divided by the rank becomes the scaling of the LoRA. Higher means stronger. A good standard value is twice your Rank.')
batch_size = gr.Slider(visible= False, label='Batch Size', value=0, minimum=0, maximum=1024, step=4, info='Now Replaced with Gradient accumulation. Keeping it for sake of old saved data')
micro_batch_size = gr.Slider(label='True Batch Size', value=4, minimum=1, maximum=128, step=1, info='Specifies how many text blocks per step will be trained. The higher value, the better the concept of training will be, but it requires more GPU memory and it reduces speed.')
grad_accumulation = gr.Slider(label='Gradient Accumulation Steps', value=1, minimum=1, maximum=256, step=1, info="Virtually multiplies the Batch Size by averaging the learning over more than one step. VRAM friendly. Evens out loss fluctuations but can also degrade training fidelity.")
with gr.Column():
stop_at_loss = gr.Slider(label='Stop at loss (Can be changed during training)', minimum=0.0, maximum=3.0, step=0.1, value=0.00, info='The process will automatically stop once the desired loss value is reached.')
gr.Markdown(" ")
epochs = gr.Number(label='Epochs', value=3, info='Number of times every entry in the dataset should be fed into training. So 1 means feed each item in once, 5 means feed it in five times, etc.')
learning_rate = gr.Textbox(label='Learning Rate', value='3e-4', info='In scientific notation. 3e-4 is a good starting base point. 1e-2 is extremely high, 1e-6 is extremely low.')
lr_scheduler_type = gr.Dropdown(label='LR Scheduler', value='linear', choices=['linear', 'constant', 'constant_with_warmup', 'cosine', 'cosine_with_restarts', 'polynomial', 'inverse_sqrt', 'FP_low_epoch_annealing', 'FP_half_time_annealing','FP_raise_fall_creative'], info='Learning rate scheduler - defines how the learning rate changes over time. Custom schedulers: FP_low_epoch_annealing, FP_half_time_annealing, FP_raise_fall_creative (see README)', elem_classes=['slim-dropdown'])
with gr.Accordion(label='Checkpoints', open=True):
with gr.Row():
with gr.Column():
save_steps = gr.Number(label='Save every n steps', value=0, info='A checkpoint will be saved every n steps and at each Epoch boundary. (0 = OFF)')
with gr.Column():
save_steps_under_loss = gr.Slider(label='Save at 10% Loss change', value=1.8, minimum=0.0, maximum=3.0, step=0.1, info="Saves checkpoints at (or bellow) this loss and then each time loss falls by at least 10% This works independently from 'Save every n steps'")
with gr.Row():
save_chackpoint_now = gr.Button('Queue Checkpoint Now')
with gr.Accordion(label='Advanced Options', open=True):
with gr.Row():
with gr.Column():
warmup_steps = gr.Number(label='Warmup Steps', value=100, info='Number of max steps used for a linear warmup. Reduces early over-fitting by the first training blocks. Value has precedent over Warmup Ratio. Aligns to the closest multiple of graddient accumulation')
warmup_ratio = gr.Slider(label='Warmup Ratio', minimum=0.0, maximum=0.2, step=0.025, value=0.0, info='Ratio of total training steps that will be used for a linear warmup. It applies only if Warmup Step is 0.')
neft_noise_alpha = gr.Slider(label='NEFtune noise scale', minimum=0.0, maximum=15, step=1, value=0.0, info='Add noise to the training to improve generalization. [0 - OFF, Starting value to experiment: 5]')
training_projection = gr.Radio(value = train_choices[4], label='LLaMA Target Projections', info='Change the targets (LORA is typically q-v)', choices=train_choices)
lora_dropout = gr.Slider(label='LoRA Dropout', minimum=0.0, maximum=1.0, step=0.025, value=0.05, info='Percentage probability for dropout of LoRA layers. This can help reduce overfitting. Most users should leave at default.')
optimizer = gr.Dropdown(label='Optimizer', value='adamw_torch', choices=['adamw_hf', 'adamw_torch', 'adamw_torch_fused', 'adamw_torch_xla', 'adamw_apex_fused', 'adafactor', 'adamw_bnb_8bit', 'adamw_anyprecision', 'sgd', 'adagrad'], info='Different optimizer implementation options, for advanced users. Effects of different options are not well documented yet.', elem_classes=['slim-dropdown'])
with gr.Column():
train_only_after = gr.Textbox(label='Train Only After', value='', info='Only consider text *after* this string in any given chunk for training. For Alpaca datasets, use "### Response:" to only train the response and ignore the input.')
add_bos_token = gr.Checkbox(label='Add BOS token', value=True, info="Adds BOS token for each dataset item")
add_eos_token = gr.Checkbox(label='Add EOS token', value=False, info="Adds EOS token for each dataset item")
add_eos_token_type = gr.Dropdown(label='EOS placement (Text file)', choices=['Every Block', 'Hard Cut Blocks Only'], value='Every Block', info='', allow_custom_value = False)
higher_rank_limit = gr.Checkbox(label='Enable higher ranks', value=False, info='If checked, changes Rank/Alpha slider above to go much higher. This will not work without a datacenter-class GPU.')
report_to = gr.Radio(label="Save detailed logs with", value="None", choices=["None", "wandb", "tensorboard"], interactive=True)
# for future
#with gr.Accordion(label='Dynamic Scheduler', open = False):
# ds_min_epochs = gr.Number(label='Minimum Epochs', value='1', info='Minimum epochs that will be always performed before ramp down can be triggered')
# ds_max_epochs = gr.Number(label='Maximum Epochs (fallback)', value='50', info='Maximum Epochs before the training will bail out completely (should be a large number)')
# ds_loss_trigger = gr.Slider(label='Trigger Loss', minimum=0.0, maximum=2.8, step=0.1, value=1.6, info='Loss at which the ramp down schedule will be triggered')
# ds_loss_rolling_window = gr.Number(label='Loss rolling average', value='4', info='Calculate loss by averaging last x numbers to avoid jumps and noise')
# ds_epochs_to_ramp = gr.Slider(label='Ramp down ratio', minimum=0.0, maximum=2.0, step=0.1, value=1.00, info='How long the ramp down will last relative to ellapsed steps (before trigger)')
# gr.Markdown('These are settings for FP_dynamic_loss_trigger scheduler. The scheduler will do warm up, then hold constant untill a loss falls under Trigger Loss, then it will commence linear ramp down schedule and stop. The length of ramp down is set by Ramp down ratio where (ramp down steps) = ratio * (elapsed steps). (The time to completition shown will be very high untill ramp down is triggered.)')
with gr.Column():
with gr.Tab(label='Formatted Dataset'):
with gr.Row():
with gr.Column():
with gr.Row():
dataset = gr.Dropdown(choices=get_datasets('training/datasets', 'json'), value='None', label='Dataset', info='The dataset file to use for training.', elem_classes=['slim-dropdown'])
create_refresh_button(dataset, lambda: None, lambda: {'choices': get_datasets('training/datasets', 'json')}, 'refresh-button')
with gr.Row():
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.', elem_classes=['slim-dropdown'])
create_refresh_button(eval_dataset, lambda: None, lambda: {'choices': get_datasets('training/datasets', 'json')}, 'refresh-button')
with gr.Column():
with gr.Row():
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.', elem_classes=['slim-dropdown'])
create_refresh_button(format, lambda: None, lambda: {'choices': get_datasets('training/formats', 'json')}, 'refresh-button')
with gr.Row():
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="Text file"):
with gr.Row():
raw_text_file = gr.Dropdown(choices=get_datasets('training/datasets', 'txt'), value='None', label='Text file', info='The text file to use for training.', elem_classes=['slim-dropdown'])
create_refresh_button(raw_text_file, lambda: None, lambda: {'choices': get_datasets('training/datasets', 'txt')}, 'refresh-button')
with gr.Row():
with gr.Column():
precize_slicing_overlap = gr.Checkbox(label='Add Overlapping blocks', value = True)
sliding_window = gr.Checkbox(label='DEMENTOR Long-form Learning by FP (Highly Experimental, use low epochs)', value = False, info='Deep Memorization Enforcement Through Overlapping and Repetition. (I named it, so shush). Special process for learning long-form text using low amount of epochs.')
#debug_slicer = gr.Checkbox(label='Dump sentencelist.json to logs', value = non_serialized_params['debug_slicer'], info='Debug Slicer')
with gr.Column():
hard_cut_string = gr.Textbox(label='Hard Cut String', value='\\n\\n\\n', info='String that indicates a cut between logical blocks of text (ex. Ideas or Chapters). Helps prevent unwanted overlap between unrelated ideas.')
min_chars = gr.Number(label='Ignore small blocks', value=0, info='Ignore Text blocks that have less or equal characters than this number.')
with gr.Tab(label="URL"):
with gr.Row():
with gr.Column():
download_file_url = gr.Textbox(label='Download JSON or txt file to datasets (or formats) folder', value='',info='The URL of a file to download. If on github, make sure you get url of the raw file (https://raw.githubusercontent.com/...). If huggin face, make sure the url has /resolve/ in it not /blob/')
with gr.Row():
download_check_overwrite = gr.Checkbox(label='Overwrite', value=False, info='Overwrite if file exist')
download_folder = gr.Radio(label="Destination", value='training/datasets', choices=['training/datasets', 'training/formats'], interactive=True)
download_button = gr.Button('Download')
download_status = gr.Textbox(label='Download Status', value='', interactive=False)
with gr.Row():
with gr.Column():
with gr.Row():
cutoff_len = gr.Slider(label='Chunk Length (Cutoff Length)', minimum=32, maximum=2048, value=256, step=32, info='The maximum length of a chunk (in tokens). Applies to both JSON dataset and text files. Higher values require much more VRAM.')
with gr.Row():
with gr.Column():
check_dataset_btn = gr.Button('Verify Dataset/Text File and suggest data entries')
check_dataset_txt = gr.Textbox(label='Dataset info', value='')
with gr.Row():
start_button = gr.Button("Start LoRA Training", variant='primary')
stop_button = gr.Button("Interrupt")
with gr.Accordion(label="Graph", open=True):
with gr.Row():
# show_actions_button = False - we use old gradio
plot_graph = gr.LinePlot(x="epoch", y="value", title="Loss Metrics", overlay_point=True, tooltip=["epoch", "value"], x_lim=[0, 1], y_lim=[0, 3.5], width=500, height=250)
output = gr.Markdown(value="Ready")
with gr.Tab('Perplexity evaluation', elem_id='evaluate-tab'):
with gr.Row():
with gr.Column():
models = gr.Dropdown(utils.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 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():
with gr.Column():
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.')
with gr.Column():
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.')
with gr.Row():
start_current_evaluation = gr.Button("Evaluate loaded model")
start_evaluation = gr.Button("Evaluate selected models")
stop_evaluation = gr.Button("Interrupt")
with gr.Column():
evaluation_log = gr.Markdown(value='')
evaluation_table = gr.Dataframe(value=generate_markdown_table(), interactive=True)
with gr.Row():
save_comments = gr.Button('Save comments', elem_classes="small-button")
refresh_table = gr.Button('Refresh the table', elem_classes="small-button")
# Training events
all_params = [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, higher_rank_limit, warmup_steps, optimizer, hard_cut_string, train_only_after, stop_at_loss, add_eos_token, min_chars, report_to, precize_slicing_overlap, add_eos_token_type, save_steps_under_loss, add_bos_token, training_projection,sliding_window,warmup_ratio,grad_accumulation, neft_noise_alpha]
def fix_old_version(batch_size_val,micro_batch_size_val, grad_accumulation_val):
if batch_size_val>0:
gradient_acc = batch_size_val // micro_batch_size_val
print(f"Using Old version of Batch Size ({batch_size_val}) to set Gradient Accumulation: {gradient_acc}")
return gradient_acc
return grad_accumulation_val
copy_from.change(partial(do_copy_params, all_params= all_params), copy_from, all_params).then(fix_old_version,[batch_size,micro_batch_size, grad_accumulation],grad_accumulation)
start_button.click(do_train, all_params, [output,plot_graph])
stop_button.click(do_interrupt, None, None, queue=False)
higher_rank_limit.change(change_rank_limit, [higher_rank_limit], [lora_rank, lora_alpha])
def trigger_stop_at_loss(stop_at_loss_value):
non_serialized_params.update({"stop_at_loss": stop_at_loss_value})
if non_serialized_params['training_loop']:
print(f"Queue: [Stop at loss Change] to {stop_at_loss_value}")
stop_at_loss.change(trigger_stop_at_loss, stop_at_loss, None)
def trigger_save_checkpoint():
non_serialized_params.update({"save_checkpoint_now": True})
if non_serialized_params['training_loop']:
print("Queue: [Save checkpoint] Checkpoint will be saved after the current step is finished.")
else:
print("Use during the training to save the checkpoint at any time.")
def update_button():
return gr.Button.update('[Checkpoint in Queue]', variant='stop', interactive=True)
def update_button2():
time.sleep(1.0)
return gr.Button.update('Queue Checkpoint Now', variant='secondary',interactive = True)
save_chackpoint_now.click(trigger_save_checkpoint, None, None).then(update_button, None,save_chackpoint_now).then(update_button2, None,save_chackpoint_now)
dataset_calc_params = [save_steps,micro_batch_size, epochs, cutoff_len, dataset, format, raw_text_file, warmup_steps, hard_cut_string, min_chars, precize_slicing_overlap,sliding_window,warmup_ratio,grad_accumulation]
def check_dataset(save_steps:int, micro_batch_size: int, epochs: int, cutoff_len: int, dataset:str, format:str, raw_text_file:str, warmup_steps:int, hard_cut_string:str, min_chars:int, precize_slicing_overlap:bool,sliding_window:bool,warmup_ratio:float,grad_accumulation:int):
result = "Specify JSON dastaset or Text file"
total_blocks = 0
if shared.tokenizer is None:
yield "Tokenizer is not available. Please Load some Model first."
return
if raw_text_file not in ['None', '']:
logger.info("Loading Text file...")
fullpath = clean_path('training/datasets', f'{raw_text_file}')
fullpath = Path(fullpath)
if fullpath.is_dir():
logger.info('Training path directory {}'.format(raw_text_file))
raw_text = ""
file_paths = sorted(fullpath.glob('*.txt'), key=lambda path: natural_keys(path.name))
for file_path in file_paths:
if file_path.is_file():
with file_path.open('r', encoding='utf-8') as file:
raw_text += file.read().replace('\r', '')
logger.info(f"Loaded training file: {file_path.name}")
else:
try:
with open(clean_path('training/datasets', f'{raw_text_file}.txt'), 'r', encoding='utf-8') as file:
raw_text = file.read().replace('\r', '')
except:
yield f"{raw_text_file}.txt doesn't seem to exsist anymore... check your training/datasets folder"
return
if min_chars<0:
min_chars = 0
# == New more precise slicing on sentence boundary ==
if sliding_window:
text_chunks = sliding_block_cut(raw_text, min_chars, False, cutoff_len, hard_cut_string,non_serialized_params['debug_slicer'])
else:
text_chunks = precise_cut(raw_text, precize_slicing_overlap, min_chars, False, cutoff_len, hard_cut_string,non_serialized_params['debug_slicer'])
total_blocks = len(text_chunks)
result = f"Text: ({raw_text_file}.txt) has {total_blocks} blocks (Block Size {cutoff_len} tokens)"
del text_chunks
else:
if dataset in ['None', '']:
yield "Select dataset or text file."
return
if format in ['None', '']:
yield "Select format choice for dataset."
return
with open(clean_path('training/formats', f'{format}.json'), 'r', encoding='utf-8-sig') as formatFile:
format_data: dict[str, str] = json.load(formatFile)
def generate_prompt(data_point: dict[str, str]):
for options, data in format_data.items():
if set(options.split(',')) == set(x[0] for x in data_point.items() if (type(x[1]) is str and len(x[1].strip()) > 0)):
for key, val in data_point.items():
if type(val) is str:
data = data.replace(f'%{key}%', val)
return data
raise RuntimeError(f'Data-point "{data_point}" has no keyset match within format "{list(format_data.keys())}"')
def tokenize_dummy(prompt):
input_ids = shared.tokenizer.encode(prompt, truncation=True, max_length=cutoff_len)
labels = [1] * len(input_ids)
input_ids = torch.tensor(input_ids)
return {
"input_ids": input_ids,
"labels": labels,
"attention_mask": input_ids.ne(shared.tokenizer.pad_token_id),
}
def generate_and_tokenize_prompt(data_point):
prompt = generate_prompt(data_point)
return tokenize_dummy(prompt)
logger.info("Loading JSON datasets...")
data = load_dataset("json", data_files=clean_path('training/datasets', f'{dataset}.json'))
data_keys = []
if data:
if 'train' in data: # Check if the 'train' split exists in the dataset
data_keys = list(data['train'][0].keys())
print("Data Keys:", data_keys)
else:
print("The dataset is empty.")
train_data = data['train'].map(generate_and_tokenize_prompt, new_fingerprint='%030x' % random.randrange(16**30))
total_blocks = train_data.num_rows
result = f"Dataset: ({dataset}.json) has {total_blocks} blocks @ length = {cutoff_len} tokens\n(Keys: {data_keys} - Format: {format}.json): "
#for options, data in format_data.items():
# format_keys = options.split(',')
# result += f"{format_keys}, "
#result = result.rstrip()
#result = result.rstrip(',')
if total_blocks>0:
number_ofSteps = int(math.ceil(total_blocks / micro_batch_size) * epochs)
num_stepsPer_epoch = int(math.ceil(number_ofSteps/epochs))
min_warm = math.ceil(100 / grad_accumulation)
warmup_steps_suggest = min(int(min_warm*grad_accumulation), int(math.ceil(number_ofSteps * 0.1)))
warmup_steps_suggest = min(warmup_steps_suggest,num_stepsPer_epoch)
save_each_n_min = int(math.ceil(number_ofSteps/10))
save_each_n_max = int(math.ceil(number_ofSteps/5))
gradient_accumulation_max = int(total_blocks)//micro_batch_size
result += f"\n[Batch Size: {micro_batch_size}, Epochs: {epochs}, Gradient Accumulation: {grad_accumulation}]\n"
result += f"Total number of steps: {number_ofSteps}\n"
result += f"Steps per each Epoch: {num_stepsPer_epoch}\n"
result += f"Suggestions:\n"
result += f"Checkpoints: Save every {save_each_n_min} - {save_each_n_max} steps (Current: {int(save_steps)})\n"
result += f"Warmup steps: {warmup_steps_suggest} (Current: {int(warmup_steps)})"
if gradient_accumulation_max < grad_accumulation:
result += f"\n\nWARNING: Gradient Accumulation {grad_accumulation} is too high: It should be below {gradient_accumulation_max}"
yield result
return
check_dataset_btn.click(check_dataset, dataset_calc_params ,check_dataset_txt)
# Evaluation events. For some reason, the interrupt event
# doesn't work with the .then() syntax, so I write them one
# by one in this ugly but functional way.
ev = start_evaluation.click(calculate_perplexity, [models, evaluate_text_file, stride_length, max_length], evaluation_log, show_progress=False)
start_evaluation.click(generate_markdown_table, None, evaluation_table, show_progress=False)
start_current_evaluation.click(lambda: ['current model'], None, tmp)
ev_cur = start_current_evaluation.click(calculate_perplexity, [tmp, evaluate_text_file, stride_length, max_length], evaluation_log, show_progress=False)
start_current_evaluation.click(generate_markdown_table, None, evaluation_table, show_progress=False)
stop_evaluation.click(None, None, None, cancels=[ev, ev_cur], queue=False)
refresh_table.click(generate_markdown_table, None, evaluation_table, show_progress=True)
save_comments.click(
save_past_evaluations, evaluation_table, None).then(
lambda: "Comments saved.", None, evaluation_log, show_progress=False)
def reload_lora():
return gr.Dropdown.update(choices=get_available_loras_local(non_serialized_params['Lora_sortedByTime']))
# nonserialized items
sort_byTime.change(lambda x: non_serialized_params.update({"Lora_sortedByTime": x}), sort_byTime, None).then(reload_lora,None,copy_from)
#debug_slicer.change(lambda x: non_serialized_params.update({"debug_slicer": x}), debug_slicer, None)
def update_dataset():
return gr.update(choices=get_datasets('training/datasets', 'json')), gr.update(choices=get_datasets('training/datasets', 'txt'))
download_button.click(download_file_from_url, [download_file_url,download_check_overwrite,download_folder] , download_status).then(update_dataset,None,[dataset , raw_text_file])
def get_datasets(path: str, ext: str):
# include subdirectories for raw txt files to allow training from a subdirectory of txt files
#if ext == "txt":
# return ['None'] + sorted(set([k.stem for k in list(Path(path).glob('txt')) + list(Path(path).glob('*/')) if k.stem != 'put-trainer-datasets-here']), key=natural_keys)
return ['None'] + sorted(set([k.stem for k in Path(path).glob(f'*.{ext}') if k.stem != 'put-trainer-datasets-here']), key=natural_keys)
def do_interrupt():
global WANT_INTERRUPT
WANT_INTERRUPT = True
def do_copy_params(lora_name: str, all_params):
if lora_name:
f_name = f"{shared.args.lora_dir}/{clean_path(None, lora_name)}/training_parameters.json"
if Path(f_name).is_file():
with open(f_name, 'r', encoding='utf-8') as format_file:
params: dict[str, str] = json.load(format_file)
else:
params = {}
else:
params = {}
result = list()
for i in range(0, len(PARAMETERS)):
key = PARAMETERS[i]
if key in params:
result.append(params[key])
else:
result.append(all_params[i])
return result
def change_rank_limit(use_higher_ranks: bool):
mult = 2 if use_higher_ranks else 1
return {"maximum": 1024 * mult, "__type__": "update"}, {"maximum": 2048 * mult, "__type__": "update"}
def clean_path(base_path: str, path: str):
"""Strips unusual symbols and forcibly builds a path as relative to the intended directory."""
path = path.replace('\\', '/').replace('..', '_')
if base_path is None:
return path
return f'{Path(base_path).absolute()}/{path}'
def backup_adapter(input_folder):
# Get the creation date of the file adapter_model.bin
try:
adapter_file = Path(f"{input_folder}/adapter_model.bin")
if adapter_file.is_file():
logger.info("Backing up existing LoRA adapter...")
creation_date = datetime.fromtimestamp(adapter_file.stat().st_ctime)
creation_date_str = creation_date.strftime("Backup-%Y-%m-%d")
# Create the new subfolder
subfolder_path = Path(f"{input_folder}/{creation_date_str}")
subfolder_path.mkdir(parents=True, exist_ok=True)
# Check if the file already exists in the subfolder
backup_adapter_file = Path(f"{input_folder}/{creation_date_str}/adapter_model.bin")
if backup_adapter_file.is_file():
print(" - Backup already exists. Skipping backup process.")
return
# Copy existing files to the new subfolder
existing_files = Path(input_folder).iterdir()
for file in existing_files:
if file.is_file():
shutil.copy2(file, subfolder_path)
except Exception as e:
print("An error occurred in backup_adapter:", str(e))
def calc_trainable_parameters(model):
trainable_params = 0
all_param = 0
for _, param in model.named_parameters():
num_params = param.numel()
# if using DS Zero 3 and the weights are initialized empty
if num_params == 0 and hasattr(param, "ds_numel"):
num_params = param.ds_numel
all_param += num_params
if param.requires_grad:
trainable_params += num_params
return trainable_params, all_param
def do_train(lora_name: str, always_override: bool, save_steps: int, micro_batch_size: int, batch_size: int, epochs: int, learning_rate: str, lr_scheduler_type: str, lora_rank: int, lora_alpha: int, lora_dropout: float, cutoff_len: int, dataset: str, eval_dataset: str, format: str, eval_steps: int, raw_text_file: str, higher_rank_limit: bool, warmup_steps: int, optimizer: str, hard_cut_string: str, train_only_after: str, stop_at_loss: float, add_eos_token: bool, min_chars: int, report_to: str, precize_slicing_overlap: bool, add_eos_token_type: str, save_steps_under_loss: float, add_bos_token: bool, training_projection: str,sliding_window:bool,warmup_ratio:float, grad_accumulation: int,neft_noise_alpha:float):
if shared.args.monkey_patch:
from alpaca_lora_4bit.monkeypatch.peft_tuners_lora_monkey_patch import (
replace_peft_model_with_int4_lora_model
)
replace_peft_model_with_int4_lora_model()
global train_log_graph
global WANT_INTERRUPT
WANT_INTERRUPT = False
statistics['loss'] = []
statistics['loss'].append({'epoch': 0, 'value': 0})
zero_pd = pd.DataFrame(statistics['loss'])
# == Input validation / processing ==
yield "Preparing the input...", zero_pd
lora_file_path = clean_path(None, lora_name)
if lora_file_path.strip() == '':
yield "Missing or invalid LoRA file name input.", zero_pd
return
lora_file_path = f"{Path(shared.args.lora_dir)}/{lora_file_path}"
actual_lr = float(learning_rate)
model_type = type(shared.model).__name__
if model_type in MODEL_CLASSES:
model_id = MODEL_CLASSES[model_type]
else:
model_id = "llama"
if model_type == "PeftModelForCausalLM":
if len(shared.lora_names) > 0:
yield "You are trying to train a LoRA while you already have another LoRA loaded. This will work, but may have unexpected effects. *(Will continue anyway in 5 seconds, press `Interrupt` to stop.)*", zero_pd
logger.warning("Training LoRA over top of another LoRA. May have unexpected effects.")
else:
yield "Model ID not matched due to LoRA loading. Consider reloading base model. *(Will continue anyway in 5 seconds, press `Interrupt` to stop.)*", zero_pd
logger.warning("Model ID not matched due to LoRA loading. Consider reloading base model.")
else:
yield "LoRA training has only currently been validated for LLaMA, OPT, GPT-J, and GPT-NeoX models. Unexpected errors may follow. *(Will continue anyway in 5 seconds, press `Interrupt` to stop.)*", zero_pd
logger.warning(f"LoRA training has only currently been validated for LLaMA, OPT, GPT-J, and GPT-NeoX models. (Found model type: {model_type})")
time.sleep(5)
if shared.args.loader == 'GPTQ-for-LLaMa' and not shared.args.monkey_patch:
yield "LoRA training with GPTQ-for-LLaMa requires loading with `--monkey-patch`", zero_pd
return
if cutoff_len <= 0 or micro_batch_size <= 0 or actual_lr <= 0 or lora_rank <= 0 or lora_alpha <= 0:
yield "Cannot input zeroes.", zero_pd
return
#in new version we dumped this in favor of grad_accumulation
#set it to zero fo new save
batch_size = 0
gradient_accumulation_steps = grad_accumulation #batch_size // micro_batch_size
shared.tokenizer.pad_token_id = 0
shared.tokenizer.padding_side = "left"
def encode(text, prepend_bos_token):
result = shared.tokenizer.encode(text, truncation=True, max_length=cutoff_len)
# Check if the first two tokens are BOS
if len(result) >= 2 and result[:2] == [shared.tokenizer.bos_token_id, shared.tokenizer.bos_token_id]:
result = result[1:]
if not prepend_bos_token and result[0] == shared.tokenizer.bos_token_id:
result = result[1:]
return result
def tokenize(prompt, append_eos_token=False, prepend_bos_token = False):
if train_only_after == '' or train_only_after not in prompt:
input_ids = encode(prompt, prepend_bos_token)
if append_eos_token and input_ids[-1] != shared.tokenizer.eos_token_id and len(input_ids) < cutoff_len:
input_ids.append(shared.tokenizer.eos_token_id)
input_ids = [shared.tokenizer.pad_token_id] * (cutoff_len - len(input_ids)) + input_ids
labels = [1] * len(input_ids)
else:
ind = prompt.index(train_only_after) + len(train_only_after)
before_tokens = encode(prompt[:ind], prepend_bos_token)
after_tokens = encode(prompt[ind:], False)
if append_eos_token and after_tokens[-1] != shared.tokenizer.eos_token_id:
after_tokens.append(shared.tokenizer.eos_token_id)
full_length = len(after_tokens) + len(before_tokens)
if full_length > cutoff_len:
after_tokens = after_tokens[:cutoff_len - len(before_tokens)]
else:
before_tokens = [shared.tokenizer.pad_token_id] * (cutoff_len - full_length) + before_tokens
input_ids = before_tokens + after_tokens
labels = [-100] * len(before_tokens) + [1] * len(after_tokens)
input_ids = torch.tensor(input_ids)
return {
"input_ids": input_ids,
"labels": labels,
"attention_mask": input_ids.ne(shared.tokenizer.pad_token_id),
}
train_template.clear()
#reset stuff
print(f"*** LoRA: {lora_name} ***")
non_serialized_params.update({"stop_at_loss": stop_at_loss})
non_serialized_params.update({"save_steps_under_loss": save_steps_under_loss+0.01})
non_serialized_params.update({"save_checkpoint_now": False})
non_serialized_params.update({"training_loop": False})
non_serialized_params.update({"current_stability": 0})
non_serialized_params.update({"save_epochs": 0})
non_serialized_params.update({"checkpoint_offset": 0})
non_serialized_params.update({"epoch_offset": 0})
train_log_graph.clear()
# === once fixed, this can be removed ==============================
if hasattr(torch.utils.checkpoint, 'noop_context_fn'):
print("Testing Pytorch...")
old_checkpoint_signature = inspect.signature(torch.utils.checkpoint.checkpoint)
# Get the signature of your new checkpoint function
my_checkpoint_signature = inspect.signature(my_checkpoint)
# Check if the signatures match
if old_checkpoint_signature.parameters == my_checkpoint_signature.parameters:
print(F"{RED}Overriding Torch checkpoint function to avoid repeated 'use_reentrant not explicitly set' warnings{RESET}")
#print(" - Note: Transformers need to pass use_reentrant in llama.modeling_llama in def forward, layer_outputs = torch.utils.checkpoint.checkpoint")
#print(" Once they do, this function can be removed")
torch.utils.checkpoint.checkpoint = my_checkpoint
# END OF FPHAM SENTENCE SPLIT functions ===================
# == Prep the dataset, format, etc ==
if raw_text_file not in ['None', '']:
train_template["template_type"] = "raw_text"
logger.info("Loading text file...")
fullpath = clean_path('training/datasets', f'{raw_text_file}')
fullpath = Path(fullpath)
if fullpath.is_dir():
logger.info('Training path directory {}'.format(raw_text_file))
raw_text = ""
file_paths = sorted(fullpath.glob('*.txt'), key=lambda path: natural_keys(path.name))
for file_path in file_paths:
if file_path.is_file():
with file_path.open('r', encoding='utf-8') as file:
raw_text += file.read().replace('\r', '')
logger.info(f"Loaded training file: {file_path.name}")
else:
with open(clean_path('training/datasets', f'{raw_text_file}.txt'), 'r', encoding='utf-8') as file:
raw_text = file.read().replace('\r', '')
# FPHAM PRECISE SLICING
if min_chars<0:
min_chars = 0
add_EOS_to_all = add_eos_token and add_eos_token_type == 'Every Block'
add_EOS_to_HC = add_eos_token and add_eos_token_type != 'Every Block'
#print (f"add_eos_token {add_eos_token}, add_EOS_to_all {add_EOS_to_all}, add_EOS_to_HC {add_EOS_to_HC}")
# == New more precise slicing on sentence boundary ==
if sliding_window:
text_chunks = sliding_block_cut(raw_text, min_chars, add_EOS_to_HC, cutoff_len, hard_cut_string,non_serialized_params['debug_slicer'])
else:
text_chunks = precise_cut(raw_text, precize_slicing_overlap, min_chars, add_EOS_to_HC, cutoff_len, hard_cut_string,non_serialized_params['debug_slicer'])
train_data = Dataset.from_list([tokenize(x, add_EOS_to_all, add_bos_token) for x in text_chunks])
if add_EOS_to_all:
print(f"Added EOS to {len(text_chunks)} blocks")
print(f"All Data Blocks: {len(text_chunks)}")
del text_chunks
eval_data = None
else:
if dataset in ['None', '']:
yield "Missing dataset choice input, cannot continue.", zero_pd
return
if format in ['None', '']:
yield "Missing format choice input, cannot continue.", zero_pd
return
train_template["template_type"] = "dataset"
with open(clean_path('training/formats', f'{format}.json'), 'r', encoding='utf-8-sig') as formatFile:
format_data: dict[str, str] = json.load(formatFile)
# == store training prompt ==
for _, value in format_data.items():
prompt_key = f"template_{len(train_template)}"
train_template[prompt_key] = value
def generate_prompt(data_point: dict[str, str]):
for options, data in format_data.items():
if set(options.split(',')) == set(x[0] for x in data_point.items() if (type(x[1]) is str and len(x[1].strip()) > 0)):
for key, val in data_point.items():
if type(val) is str:
data = data.replace(f'%{key}%', val)
return data
raise RuntimeError(f'Data-point "{data_point}" has no keyset match within format "{list(format_data.keys())}"')
def generate_and_tokenize_prompt(data_point):
prompt = generate_prompt(data_point)
return tokenize(prompt, add_eos_token, add_bos_token)
logger.info("Loading JSON datasets...")
data = load_dataset("json", data_files=clean_path('training/datasets', f'{dataset}.json'))
train_data = data['train'].map(generate_and_tokenize_prompt, new_fingerprint='%030x' % random.randrange(16**30))
print(f"BOS: {add_bos_token} EOS: {add_eos_token}")
print(f"Data Blocks: {train_data.num_rows}")
if eval_dataset == 'None':
eval_data = None
else:
eval_data = load_dataset("json", data_files=clean_path('training/datasets', f'{eval_dataset}.json'))
eval_data = eval_data['train'].map(generate_and_tokenize_prompt, new_fingerprint='%030x' % random.randrange(16**30))
# == We MUST reload model if it went through any previous training, even failed one ==
if shared.model_dirty_from_training:
selected_model = shared.model_name
if selected_model:
print("\033[1;31;1m(Model has been modified by previous training, it needs to be reloaded...)\033[0;37;0m")
try:
yield f"Reloading {selected_model}...", zero_pd
reload_model()
shared.tokenizer.pad_token_id = 0
shared.tokenizer.padding_side = "left"
if shared.model is not None:
print("Model reloaded OK, continue with training.")
else:
return f"Failed to load {selected_model}."
except:
exc = traceback.format_exc()
logger.error('Failed to reload the model.')
print(exc)
return exc.replace('\n', '\n\n')
# == Start prepping the model itself ==
if not hasattr(shared.model, 'lm_head') or hasattr(shared.model.lm_head, 'weight'):
logger.info("Getting model ready...")
# here we can disable gradient checkpoint, by default = true, use_gradient_checkpointing=True
prepare_model_for_kbit_training(shared.model)
# base model is now frozen and should not be reused for any other LoRA training than this one
shared.model_dirty_from_training = True
print(f"Transformers Model Type: {YELLOW}{model_type}{RESET}")
if training_projection==train_choices[0]:
model_to_lora_modules[model_id] = ["gate_proj","down_proj","up_proj","q_proj","k_proj","v_proj","o_proj"]
elif training_projection==train_choices[1]:
model_to_lora_modules[model_id] = ["q_proj","k_proj", "v_proj", "o_proj"]
elif training_projection==train_choices[2]:
model_to_lora_modules[model_id] = ["q_proj","k_proj", "v_proj"]
elif training_projection==train_choices[3]:
model_to_lora_modules[model_id] = ["k_proj", "v_proj", "down_proj"]
else:
model_to_lora_modules[model_id] = ["q_proj", "v_proj"]
logger.info("Preparing for training...")
config = LoraConfig(
r=lora_rank,
lora_alpha=lora_alpha,
target_modules=model_to_lora_modules[model_id],
lora_dropout=lora_dropout,
bias="none",
task_type="CAUSAL_LM"
)
# == Backup the existing adapter ==
if not always_override:
backup_adapter(lora_file_path)
# == get model trainable params
model_trainable_params, model_all_params = calc_trainable_parameters(shared.model)
try:
logger.info("Creating LoRA model...")
lora_model = get_peft_model(shared.model, config)
if not always_override and Path(f"{lora_file_path}/adapter_model.bin").is_file():
logger.info("Loading existing LoRA data...")
state_dict_peft = torch.load(f"{lora_file_path}/adapter_model.bin")
set_peft_model_state_dict(lora_model, state_dict_peft)
print(f" + Continue Training on {RED}{lora_file_path}/adapter_model.bin{RESET}")
#load training_log.json if exist
if Path(f"{lora_file_path}/training_log.json").is_file():
with open(f"{lora_file_path}/training_log.json", 'r') as json_file:
json_ilog = json.load(json_file)
for key, value in json_ilog.items():
if key=='current_steps':
non_serialized_params.update({"checkpoint_offset": int(value+1)})
print(f" + Checkpoints will be saved with offset: {RED}{non_serialized_params['checkpoint_offset']}{RESET}")
if key=='epoch':
non_serialized_params.update({"epoch_offset": value})
print(f" + Epoch offset: {RED}{non_serialized_params['epoch_offset']}{RESET}")
if Path(f"{lora_file_path}/training_graph.json").is_file():
try:
with open(f"{lora_file_path}/training_graph.json", 'r') as json_file:
train_log_graph = json.load(json_file)
print(" + Training Graph loaded")
except:
print(f"Can't read training_graph")
except:
yield traceback.format_exc().replace('\n', '\n\n'), zero_pd
return
if shared.args.monkey_patch:
from alpaca_lora_4bit.autograd_4bit import Autograd4bitQuantLinear
from alpaca_lora_4bit.models import Linear4bitLt
for _, m in lora_model.named_modules():
if isinstance(m, Autograd4bitQuantLinear) or isinstance(m, Linear4bitLt):
if m.is_v1_model:
m.zeros = m.zeros.half()
m.scales = m.scales.half()
class Tracked():
def __init__(self):
self.current_steps = 0
self.max_steps = 0
self.did_save = False
tracked = Tracked()
actual_save_steps = math.ceil(save_steps / gradient_accumulation_steps)
class Callbacks(transformers.TrainerCallback):
def on_step_begin(self, args: transformers.TrainingArguments, state: transformers.TrainerState, control: transformers.TrainerControl, **kwargs):
tracked.current_steps = state.global_step * gradient_accumulation_steps
tracked.max_steps = state.max_steps * gradient_accumulation_steps
ssteps10 = int(max(2,(state.max_steps/epochs)*0.1))
if WANT_INTERRUPT:
control.should_epoch_stop = True
control.should_training_stop = True
else:
current_loss = float(train_log.get('loss', 0.0))
current_epoch_int = int(float(train_log.get('epoch', 0.0)))
force_save = False
current_steps_offset = tracked.current_steps + non_serialized_params['checkpoint_offset']
folder_save = f"checkpoint-{current_steps_offset}"
# save if triggered by user
if non_serialized_params['save_checkpoint_now']:
force_save = True
non_serialized_params.update({"save_checkpoint_now": False})
print(f"\033[1;31;1mSave Checkpoint manually trigerred.\033[0;37;0m")
folder_save = f"checkpoint-{current_steps_offset}-user"
patience = 3 # Set the number of consecutive steps for tracking stability
if gradient_accumulation_steps==1:
patience = 4
min_steps = ssteps10
# Save each time the loss is below the threshold
if current_loss < non_serialized_params['save_steps_under_loss'] and current_loss > 0 and state.global_step > min_steps:
current_stability = non_serialized_params['current_stability']
current_stability += 1
non_serialized_params.update({"current_stability": current_stability})
if current_stability >= patience:
current_stability = 0
non_serialized_params.update({"current_stability": current_stability})
current_loss_dec = round(current_loss, 2)
loss_str = f"{current_loss_dec:.2f}"
loss_str = loss_str.replace('.', '_')
new_save = (current_loss_dec-0.1) + 0.01
non_serialized_params.update({"save_steps_under_loss": new_save})
folder_save = f"checkpoint-{current_steps_offset}-loss-{loss_str}"
force_save = True
else:
# Reset stability if the loss goes above the threshold
non_serialized_params.update({"current_stability": 0})
# Save full epochs
if actual_save_steps>0 and current_epoch_int > non_serialized_params['save_epochs'] and state.global_step > min_steps:
current_epoch_offset = current_epoch_int
if non_serialized_params['epoch_offset'] > 0:
current_epoch_offset = current_epoch_int + round(non_serialized_params['epoch_offset'], 2)
ep_off_str = f"{current_epoch_offset}"
ep_off_str = ep_off_str.replace('.', '_')
folder_save = f"checkpoint-{current_steps_offset}-epoch-{ep_off_str}"
non_serialized_params.update({"save_epochs": current_epoch_int})
force_save = True
# save each actual_save_steps
if state.global_step > 0 and actual_save_steps > 0 and state.global_step % actual_save_steps == 0:
folder_save = f"checkpoint-{current_steps_offset}"
force_save = True
if force_save:
lora_model.save_pretrained(f"{lora_file_path}/{folder_save}/")
print(f"\033[1;30;40mStep: {tracked.current_steps:6} \033[0;37;0m Saved: [{folder_save}]")
# Save log
with open(f"{lora_file_path}/{folder_save}/training_log.json", 'w', encoding='utf-8') as file:
json.dump(train_log, file, indent=2)
# == Save training prompt ==
with open(f"{lora_file_path}/{folder_save}/training_prompt.json", 'w', encoding='utf-8') as file:
json.dump(train_template, file, indent=2)
def on_substep_end(self, args: transformers.TrainingArguments, state: transformers.TrainerState, control: transformers.TrainerControl, **kwargs):
tracked.current_steps += 1
if WANT_INTERRUPT:
control.should_epoch_stop = True
control.should_training_stop = True
def on_log(self, args: transformers.TrainingArguments, state: transformers.TrainerState, control: transformers.TrainerControl, logs, **kwargs):
train_log.update(logs)
current_steps_offset = tracked.current_steps + non_serialized_params['checkpoint_offset']
current_epoch_offset = train_log.get('epoch', 0.0) + non_serialized_params['epoch_offset']
train_log.update({"current_steps": tracked.current_steps})
train_log.update({"current_steps_adjusted": current_steps_offset})
train_log.update({"epoch_adjusted": current_epoch_offset})
if WANT_INTERRUPT:
print("\033[1;31;1mInterrupted by user\033[0;37;0m")
if non_serialized_params['checkpoint_offset']>0:
print(f"\033[1;30;40mStep: {tracked.current_steps:6} [+{non_serialized_params['checkpoint_offset']}] \033[0;37;0m", end='')
else:
print(f"\033[1;30;40mStep: {tracked.current_steps:6} \033[0;37;0m", end='')
graphentry = {
'current_steps': int(train_log.get('current_steps_adjusted',0)),
'loss': float(train_log.get('loss', 0.0)),
'learning_rate': float(train_log.get('learning_rate', 0.0)),
'epoch': float(train_log.get('epoch_adjusted', 0.0))
}
cur_loss = float(train_log.get('loss', 0.0))
cur_lr = float(train_log.get('learning_rate', 0.0))
cur_epoch = float(train_log.get('epoch', 0.0))
if len(statistics['loss']) == 1:
first_epoch = statistics['loss'][0]['epoch']
first_value = statistics['loss'][0]['value']
if first_value ==0:
statistics['loss'] = []
statistics['loss'].append({'epoch': cur_epoch, 'value': cur_loss})
statistics['lr'].append({'epoch': cur_epoch, 'value': cur_lr})
# Add the entry to the continuous log
train_log_graph.append(graphentry)
# Save the graph log for now, we can later generate full graph
with open(f"{lora_file_path}/training_graph.json", 'w') as file:
json.dump(train_log_graph, file, indent=4)
if 'loss' in logs:
loss = float(logs['loss'])
if loss <= stop_at_loss:
control.should_epoch_stop = True
control.should_training_stop = True
print(f"{RED}Stop Loss {stop_at_loss} reached.{RESET}")
# FPHAM SAMPLE REQ Transformers error handling
gradient_accumulation_max = int(train_data.num_rows)//micro_batch_size
if gradient_accumulation_max < gradient_accumulation_steps:
print(f"{RED}WARNING:{RESET} Current gradient accumulation is {RED}too high{RESET} for the amount of training data.")
print(f"Gradient accumulation: {gradient_accumulation_steps} should be less than: {gradient_accumulation_max}. {RED}This could crash Accelerate/Transformers{RESET}")
#min_batchSize = sample_req*micro_batch_size
print(f"Preferable fix: {RED}Increase the size of dataset{RESET}")
print(f"... or Decrerase Gradient Accumulation {RED}{gradient_accumulation_steps}{RESET} to below {GREEN}{gradient_accumulation_max}{RESET}")
gradient_accumulation_steps = max(1,gradient_accumulation_max-1)
print(f"Last resort fix for this run: Lowering Gradient accumulation to {GREEN}{gradient_accumulation_steps}{RESET} [Good luck]")
else:
print(f"Data Size Check: Gradient accumulation: {YELLOW}{gradient_accumulation_steps}{RESET} <= Blocks/Batch {gradient_accumulation_max} ... {GREEN}[OK]{RESET}")
#END OF FPHAM SAMPLE REQ
# FPHAM Custom Scheduler ==
custom_scheduller = False
lr_scheduler_type_arg = lr_scheduler_type
if lr_scheduler_type == 'FP_low_epoch_annealing':
custom_scheduller = True
lr_scheduler_type_arg = 'cosine'
elif lr_scheduler_type == 'FP_half_time_annealing':
custom_scheduller = True
lr_scheduler_type_arg = 'constant'
elif lr_scheduler_type =='FP_raise_fall_creative':
custom_scheduller = True
lr_scheduler_type_arg = 'constant_with_warmup'
#gradient_checkpointing=True
args=transformers.TrainingArguments(
report_to=report_to if report_to != "None" else None,
per_device_train_batch_size=micro_batch_size,
gradient_accumulation_steps=gradient_accumulation_steps,
warmup_steps=math.ceil(warmup_steps / gradient_accumulation_steps),
warmup_ratio = warmup_ratio,
num_train_epochs=epochs,
learning_rate=actual_lr,
fp16=False if shared.args.cpu else True,
optim=optimizer,
logging_steps=1,
evaluation_strategy="steps" if eval_data is not None else "no",
eval_steps=math.ceil(eval_steps / gradient_accumulation_steps) if eval_data is not None else None,
save_strategy="steps" if eval_data is not None else "no",
output_dir=lora_file_path,
lr_scheduler_type=lr_scheduler_type_arg,
load_best_model_at_end=eval_data is not None,
# TODO: Enable multi-device support
ddp_find_unused_parameters=None,
no_cuda=shared.args.cpu,
)
if custom_scheduller:
trainer = FPSchedulerTrainer(
neftune_noise_alpha=neft_noise_alpha,
model=lora_model,
train_dataset=train_data,
eval_dataset=eval_data,
args=args,
data_collator=transformers.DataCollatorForLanguageModeling(shared.tokenizer, mlm=False),
callbacks=list([Callbacks()])
)
elif neft_noise_alpha > 0:
trainer = FPNEFtuneTrainer(
neftune_noise_alpha=neft_noise_alpha,
model=lora_model,
train_dataset=train_data,
eval_dataset=eval_data,
args=args,
data_collator=transformers.DataCollatorForLanguageModeling(shared.tokenizer, mlm=False),
callbacks=list([Callbacks()])
)
else:
trainer = transformers.Trainer(
model=lora_model,
train_dataset=train_data,
eval_dataset=eval_data,
args=args,
data_collator=transformers.DataCollatorForLanguageModeling(shared.tokenizer, mlm=False),
callbacks=list([Callbacks()])
)
# END OF FPHAM CUSTOM SCHEDULER
lora_model.config.use_cache = False
if torch.__version__ >= "2" and sys.platform != "win32":
lora_model = torch.compile(lora_model)
# == Save parameters for reuse ==
with open(f"{lora_file_path}/training_parameters.json", 'w', encoding='utf-8') as file:
vars = locals()
json.dump({x: vars[x] for x in PARAMETERS}, file, indent=2)
# == Save training prompt ==
with open(f"{lora_file_path}/training_prompt.json", 'w', encoding='utf-8') as file:
json.dump(train_template, file, indent=2)
# == Main run and monitor loop ==
logger.info("Starting training...")
yield "Starting...", zero_pd
lora_trainable_param, lora_all_param = calc_trainable_parameters(lora_model)
projections_string = ", ".join([projection.replace("_proj", "") for projection in model_to_lora_modules[model_id]])
print(f"Training '{model_id}' model using {YELLOW}({projections_string}){RESET} projections")
if lora_all_param > 0:
print(f"Trainable params: {lora_trainable_param:,d} ({RED}{100 * lora_trainable_param / lora_all_param:.4f} %{RESET}), All params: {lora_all_param:,d} (Model: {model_all_params:,d})")
train_log.update({"base_model_name": shared.model_name})
train_log.update({"base_model_class": shared.model.__class__.__name__})
train_log.update({"base_loaded_in_4bit": getattr(lora_model, "is_loaded_in_4bit", False)})
train_log.update({"base_loaded_in_8bit": getattr(lora_model, "is_loaded_in_8bit", False)})
train_log.update({"projections": projections_string})
if non_serialized_params['checkpoint_offset'] > 0:
train_log.update({"last_run_steps_offset": non_serialized_params['checkpoint_offset']})
train_log.update({"last_run_epoch_offset": non_serialized_params['epoch_offset']})
if non_serialized_params['checkpoint_offset'] > 0:
print(f"Continue training on {RED}previous adapter{RESET} from epoch: {RED}{non_serialized_params['epoch_offset']}{RESET}")
if stop_at_loss > 0:
print(f"Monitoring loss {RED}(Auto-Stop at: {stop_at_loss}){RESET}")
if WANT_INTERRUPT:
yield "Interrupted before start.", zero_pd
return
def log_train_dataset(trainer):
decoded_entries = []
# Try to decode the entries and write the log file
try:
# Iterate over the first 10 elements in the dataset (or fewer if there are less than 10)
for i in range(min(10, len(trainer.train_dataset))):
decoded_text = shared.tokenizer.decode(trainer.train_dataset[i]['input_ids'])
decoded_entries.append({"value": decoded_text})
# Write the log file
Path('logs').mkdir(exist_ok=True)
with open(Path('logs/train_dataset_sample.json'), 'w') as json_file:
json.dump(decoded_entries, json_file, indent=4)
logger.info("Log file 'train_dataset_sample.json' created in the 'logs' directory.")
except Exception as e:
logger.error(f"Failed to create log file due to error: {e}")
def threaded_run():
log_train_dataset(trainer)
trainer.train()
# Note: save in the thread in case the gradio thread breaks (eg browser closed)
lora_model.save_pretrained(lora_file_path)
logger.info("LoRA training run is completed and saved.")
# Save log
with open(f"{lora_file_path}/training_log.json", 'w', encoding='utf-8') as file:
json.dump(train_log, file, indent=2)
thread = threading.Thread(target=threaded_run)
thread.start()
last_step = 0
start_time = time.perf_counter()
while thread.is_alive():
time.sleep(0.5)
if statistics['loss']:
max_value_dict = max(statistics['loss'], key=lambda x: x['value'])
max_value = max_value_dict['value']+0.4
first_epoch = statistics['loss'][0]['epoch']
last_epoch = statistics['loss'][-1]['epoch']
else:
max_value = 3.5
last_epoch = 0
first_epoch = 0
if WANT_INTERRUPT:
losses = gr.LinePlot.update(
value = pd.DataFrame(statistics['loss']),
x="epoch", y="value",
title="Loss Metrics",
overlay_point=True, tooltip=["epoch", "value"],
x_lim=[first_epoch,last_epoch], y_lim=[0,max_value],
width=500, height=250 )
yield "Interrupting, please wait... *(Run will stop after the current training step completes.)*", losses
elif tracked.current_steps != last_step:
last_step = tracked.current_steps
time_elapsed = time.perf_counter() - start_time
lastloss = float(train_log.get('loss', 0.0))
non_serialized_params.update({"training_loop": True})
if lastloss > 0:
lastloss_str = f", ... Current Loss: `{lastloss:.2f}`"
else:
lastloss_str = ""
if time_elapsed <= 0:
timer_info = ""
total_time_estimate = 999
else:
its = tracked.current_steps / time_elapsed
if its > 1:
timer_info = f"`{its:.2f}` it/s"
else:
timer_info = f"`{1.0/its:.2f}` s/it"
total_time_estimate = (1.0 / its) * (tracked.max_steps)
if stop_at_loss != non_serialized_params['stop_at_loss']:
stop_at_loss = non_serialized_params['stop_at_loss']
print(f"Stop at loss changed {RED}(Auto-Stop at: {stop_at_loss}){RESET}")
losses = gr.LinePlot.update(
value = pd.DataFrame(statistics['loss']),
x="epoch", y="value",
title="Loss Metrics",
overlay_point=True, tooltip=["epoch", "value"],
x_lim=[first_epoch,last_epoch], y_lim=[0,max_value],
width=500, height=250 )
yield f"Running... **{tracked.current_steps}** / **{tracked.max_steps}** ... {timer_info}, {format_time(time_elapsed)} / {format_time(total_time_estimate)} ... {format_time(total_time_estimate - time_elapsed)} remaining {lastloss_str}", losses
# Saving in the train thread might fail if an error occurs, so save here if so.
#return_pd = pd.DataFrame(statistics['loss'])
if statistics['loss']:
max_value_dict = max(statistics['loss'], key=lambda x: x['value'])
max_value = max_value_dict['value']+0.4
first_epoch = statistics['loss'][0]['epoch']
last_epoch = statistics['loss'][-1]['epoch']
else:
max_value = 3.5
last_epoch = 0
first_epoch = 0
return_pd = gr.LinePlot.update(
value = pd.DataFrame(statistics['loss']),
x="epoch", y="value",
title="Loss Metrics",
overlay_point=True, tooltip=["epoch", "value"],
x_lim=[first_epoch,last_epoch], y_lim=[0,max_value],
width=500, height=250)
non_serialized_params.update({"training_loop": False})
if not tracked.did_save:
logger.info("Training complete, saving...")
lora_model.save_pretrained(lora_file_path)
if WANT_INTERRUPT:
logger.info("Training interrupted.")
yield f"Interrupted by user. LoRA saved to `{lora_file_path}`.", return_pd
else:
logger.info("Training complete!")
yield f"Done! LoRA saved to `{lora_file_path}`.\n\nBefore testing your new LoRA, make sure to first reload the model, as it is currently dirty from training.", return_pd
create_graph(lora_file_path, lora_name)
def format_time(seconds: float):
if seconds < 120:
return f"`{seconds:.0f}` seconds"
minutes = seconds / 60
if minutes < 120:
return f"`{minutes:.0f}` minutes"
hours = minutes / 60
return f"`{hours:.0f}` hours"
Reference in New Issue View Git Blame Copy Permalink