Datasets:

adamkarvonen

/

chess_games

like 11

sort_split.py - sort by row (game) length, then split. Useful for training models like Mamba with flexible context size (sequence length), for faster training. See train_bygame.py below, though that's not directly relevant to this DATASET repo.

import pandas as pd
import pyarrow.parquet as pq
import os
import numpy as np
import math

def sort_and_split_parquet(input_file, output_dir, n_splits):
    # Load the parquet file
    df = pq.read_table(input_file).to_pandas()

    # Sort by the length of the 'tokenized' column
    df['length'] = df['tokenized'].apply(len)
    df_sorted = df.sort_values(by='length').drop(columns=['length'])

    # Calculate the number of rows per split
    total_rows = len(df_sorted)
    rows_per_split = math.ceil(total_rows / n_splits)

    print("Dataset sorted, splitting...")
    # Split and save each part
    for i in range(n_splits):
        start_row = i * rows_per_split
        end_row = min(start_row + rows_per_split, total_rows)
        split_df = df_sorted.iloc[start_row:end_row]

        # Save the split DataFrame as a parquet file
        split_file_name = f"train_{i}.parquet"
        split_df.to_parquet(os.path.join(output_dir, split_file_name))

        first_game_length = len(split_df.iloc[0]['tokenized'])
        last_game_length = len(split_df.iloc[-1]['tokenized'])
        print(f"Saved {split_file_name}... Game lengths: {first_game_length} - {last_game_length}")

# Example usage
input_file = 'data/chess/train.parquet'
output_dir = 'data/chess'
n_splits = 250

sort_and_split_parquet(input_file, output_dir, n_splits)
print("Done.")

csv2pqt.py

import pandas as pd
import pyarrow as pa
import pyarrow.parquet as pq
import numpy as np
import tiktoken
import pickle
from sklearn.model_selection import train_test_split
import random

def tokenize_game(game, stoi):
    # Remove the prefix and tokenize the game
    game_cleaned = game.split('\n\n', 1)[1] if '\n\n' in game else game
    return np.array([stoi[char] for char in game_cleaned], dtype=np.uint8)

if __name__ == "__main__":
    dataset_path = "gt1_8kElo_all.zip"
    meta_path = "data/chess/meta.pkl"
    
    # Load metadata for tokenization
    with open(meta_path, "rb") as f:
        meta = pickle.load(f)
    stoi = meta["stoi"]
    
    # Read CSV with headers
    df = pd.read_csv(dataset_path)
    print(df.iloc[random.randint(0, len(df) - 1)])

    # Tokenize games in the 'transcript' column
    df['tokenized'] = df['transcript'].apply(lambda x: tokenize_game(x, stoi))

    # Split dataset into training and validation
    train_df, val_df = train_test_split(df, test_size=0.01, random_state=42)

    # Define a function to write the DataFrame to a Parquet file with multiple rows per row group
    def write_parquet_with_row_groups(df, file_name, rows_per_group=100):
        table = pa.Table.from_pandas(df[['tokenized']])
        writer = pq.ParquetWriter(file_name, table.schema)
        for i in range(0, len(df), rows_per_group):
            writer.write_table(table.slice(i, min(rows_per_group, len(df) - i)))
        writer.close()

    write_parquet_with_row_groups(train_df, 'train_lich.parquet')
    write_parquet_with_row_groups(val_df, 'val_lich.parquet')

train_bygame.py (see https://github.com/alxndrTL/mamba.py)

import os
import time
import math
import pickle
from contextlib import nullcontext
import numpy as np
import torch
import torch.nn.functional as F
from torch.nn.parallel import DistributedDataParallel as DDP
from torch.distributed import init_process_group, destroy_process_group
from mamba_lm import MambaLM, MambaLMConfig
import pyarrow.parquet as pq
import random
from torch.utils.data import Dataset, DataLoader
import glob

# -----------------------------------------------------------------------------
# default config values designed for Mamba model training
# I/O
out_dir = 'out'
eval_interval = 2000
log_interval = 1
eval_iters = 5
eval_only = False
always_save_checkpoint = True
init_from = 'resume'  # 'scratch', 'resume', or Mamba model name
# wandb logging
wandb_log = False
wandb_project = 'mamba'
wandb_run_name = 'mamba_run'  # modify as needed
# data
dataset = 'chess'  # specify your dataset
gradient_accumulation_steps = 5 * 8
batch_size = 12
base_batch_size = batch_size
effective_batch_size = batch_size
max_seq_len = 1024 # A trianing-only parameter for controlling VRAM

# model
n_layer = 12
d_model = 768
dt_rank = 'auto'
d_state = 16
expand_factor = 2
bias = False
conv_bias = True
pscan = True
vocab_size = 32000

# optimizer settings
learning_rate = 6e-4
max_iters = 600000
weight_decay = 1e-1
beta1 = 0.9
beta2 = 0.95
grad_clip = 1.0
# learning rate decay settings
decay_lr = True
warmup_iters = 2000
lr_decay_iters = 600000
min_lr = 6e-5
# DDP settings
backend = 'nccl'
# system
device = 'cuda' if torch.cuda.is_available() else 'cpu'
dtype = 'bfloat16' if torch.cuda.is_bf16_supported() else 'float32'
compile = False  # set to True if using PyTorch 2.0
# -----------------------------------------------------------------------------

config_keys = [k for k, v in globals().items() if not k.startswith('_') and isinstance(v, (int, float, bool, str))]
exec(open('configurator.py').read())  # overrides from command line or config file
config = {k: globals()[k] for k in config_keys}  # will be useful for logging
# -----------------------------------------------------------------------------

mamba_config = MambaLMConfig(
    d_model=d_model,  # adjust as needed
    n_layers=n_layer,  # adjust as needed
    dt_rank=dt_rank,
    d_state=d_state,
    expand_factor=expand_factor,
    bias=bias,
    conv_bias=conv_bias,
    pscan=pscan,
    vocab_size=vocab_size # adjust based on your dataset
)

# DDP and other initializations
ddp = int(os.environ.get('RANK', -1)) != -1
if ddp:
    init_process_group(backend=backend)
    ddp_rank = int(os.environ['RANK'])
    ddp_local_rank = int(os.environ['LOCAL_RANK'])
    ddp_world_size = int(os.environ['WORLD_SIZE'])
    device = f'cuda:{ddp_local_rank}'
    torch.cuda.set_device(device)
    master_process = ddp_rank == 0
    seed_offset = ddp_rank
    assert gradient_accumulation_steps % ddp_world_size == 0
    gradient_accumulation_steps //= ddp_world_size
else:
    master_process = True
    seed_offset = 0
    ddp_world_size = 1
tokens_per_iter = gradient_accumulation_steps * ddp_world_size * batch_size * max_seq_len

if master_process:
    os.makedirs(out_dir, exist_ok=True)
torch.manual_seed(1337 + seed_offset)
torch.backends.cuda.matmul.allow_tf32 = True
torch.backends.cudnn.allow_tf32 = True
device_type = 'cuda' if 'cuda' in device else 'cpu'
ptdtype = {'float32': torch.float32, 'bfloat16': torch.bfloat16}[dtype]
ctx = nullcontext() if device_type == 'cpu' else torch.amp.autocast(device_type=device_type, dtype=ptdtype)

# poor man's data loader
data_dir = os.path.join('data', dataset)
current_train_file_index = 0
train_file_update_interval = eval_interval // 20
train_files = glob.glob(os.path.join(data_dir, 'train*.parquet'))
train_datasets = []
for f in train_files:
    dataset = pq.read_table(f).to_pandas()
    dataset = dataset[dataset['tokenized'].apply(len) >= 8]
    train_datasets.append(dataset)
val_data = pq.read_table(os.path.join(data_dir, 'val.parquet')).to_pandas()
val_data = val_data[val_data['tokenized'].apply(len) >= 8]
truncated_games_count = 0
total_games_count = 0
def get_batch(split):
    global truncated_games_count, total_games_count

    # Randomly select batch_size games
    dataset = train_datasets[current_train_file_index] if split == 'train' else val_data  # Use the correct DataFrame based on the split
    sample_df = dataset.sample(batch_size)
    games = sample_df['tokenized'].tolist()

    # Prepare sequences tensor for the batch
    max_length_in_batch = min(max(len(game) for game in games), max_seq_len)
    sequences = torch.zeros((batch_size, max_length_in_batch), dtype=torch.int64)

    for i, game in enumerate(games):
        total_games_count += 1

        if len(game) > max_seq_len:
            truncated_games_count += 1
            # Randomly decide truncation strategy
            truncation_choice = random.choice(['beginning', 'end', 'end2', 'random'])
            if truncation_choice == 'beginning':
                # Truncatethe beginning (use from the end backward)
                truncated_game = game[-max_seq_len:]
            elif truncation_choice.startswith('end'):
                # Truncatethe end (use from the beginning forward)
                truncated_game = game[:max_seq_len]
            else:
                # Random start index (truncate beginning and end)
                start_idx = random.randint(0, len(game) - max_seq_len)
                truncated_game = game[start_idx:start_idx + max_seq_len]
            sequences[i, :len(truncated_game)] = torch.tensor(truncated_game, dtype=torch.int64)
            # Report the percentage of truncated games
            if truncated_games_count > 0 and truncated_games_count % 50 == 0:
                truncated_percentage = (truncated_games_count / total_games_count) * 100
                print(f"Percentage of truncated games: {truncated_percentage:.2f}%\t\t({truncated_games_count}/{total_games_count})")
        else:
            sequences[i, :len(game)] = torch.tensor(game, dtype=torch.int64)
            
    if device_type == 'cuda':
        sequences = sequences.pin_memory().to(device, non_blocking=True)
    else:
        sequences = sequences.to(device)

    return sequences
    
# init these up here, can override if init_from='resume' (i.e. from a checkpoint)
iter_num = 0
best_val_loss = 1e9

# attempt to derive vocab_size from the dataset
meta_path = os.path.join(data_dir, 'meta.pkl')
meta_vocab_size = None
if os.path.exists(meta_path):
    with open(meta_path, 'rb') as f:
        meta = pickle.load(f)
    meta_vocab_size = meta['vocab_size']
    print(f"found vocab_size = {meta_vocab_size} (inside {meta_path})")
    
# Model initialization
if init_from == 'scratch':
    print("Initializing a new Mamba model from scratch")
    if meta_vocab_size is None:
        print(f"defaulting to vocab_size of {vocab_size}")
    else:
        mamba_config.vocab_size = meta_vocab_size
    model = MambaLM(mamba_config)
elif init_from == 'resume':
    print(f"Resuming training from {out_dir}")
    ckpt_path = os.path.join(out_dir, 'ckpt.pt')
    checkpoint = torch.load(ckpt_path, map_location=device)
    mamba_config = checkpoint['model_args']
    model = MambaLM(mamba_config)
    state_dict = checkpoint['model']
    # fix the keys of the state dictionary :(
    # honestly no idea how checkpoints sometimes get this prefix, have to debug more
    unwanted_prefix = '_orig_mod.'
    for k,v in list(state_dict.items()):
        if k.startswith(unwanted_prefix):
            state_dict[k[len(unwanted_prefix):]] = state_dict.pop(k)
    model.load_state_dict(state_dict)
    if 'effective_batch_size' not in checkpoint['config']:
        print("Checkpoint was saved without `effective_batch_size`, assuming current value (will save with next checkpoint). This is used for correcting `iter_num` when the effetive batch size is changed.")
        checkpoint['config']['effective_batch_size'] = effective_batch_size
    iter_num = checkpoint['iter_num'] * (checkpoint['config']['effective_batch_size'] / effective_batch_size)
    best_val_loss = checkpoint['best_val_loss']
elif init_from.startswith('state-spaces'):
    print(f"Initializing from Mamba pre-trained weights: {init_from}")
    model = from_pretrained(init_from)
    mamba_config = model.config
else:
    raise ValueError("Invalid init_from value")

model.to(device)

print(f'Model with {sum([p.numel() for p in model.parameters()])} parameters loaded.')

# Optimizer and GradScaler
optimizer = torch.optim.AdamW(model.parameters(), lr=learning_rate, weight_decay=weight_decay, betas=(beta1, beta2))
scaler = torch.cuda.amp.GradScaler(enabled=dtype == 'float16')
if init_from == 'resume':
    optimizer.load_state_dict(checkpoint['optimizer'])
checkpoint = None

# Compile the model if using PyTorch 2.0
if compile:
    print("compiling the model... (takes a ~minute)")
    model = torch.compile(model)

# Wrap model in DDP container if necessary
if ddp:
    model = DDP(model, device_ids=[ddp_local_rank])
    



@torch

	.no_grad()
def estimate_loss():
    out = {}
    model.eval()
    for split in ['train']: #['train', 'val']:
        losses = torch.zeros(eval_iters)
        for k in range(eval_iters):
            if split=='train' and k % (train_file_update_interval // 1.5) == 0:
                current_train_file_index = random.randint(0, len(train_files) - 1)
                print(f"Switched to file: {train_files[current_train_file_index]}\t\t(eval)")
            tokens = get_batch(split)  # Fetch tokens in the correct format
            logits = model(tokens[:, :-1])  # Predict next tokens (ignore last token)

            # The targets are the tokens shifted by one position
            targets = tokens[:, 1:].reshape(-1)  # Flatten targets for cross-entropy

            # Compute cross-entropy loss between logits and targets
            loss = F.cross_entropy(logits.view(-1, logits.size(-1)), targets)
            losses[k] = loss.item()

        split = 'val' # Temporary hack
        out[split] = losses.mean()
    model.train()
    return out


# learning rate decay scheduler (cosine with warmup)
def get_lr(it):
    # 1) linear warmup for warmup_iters steps
    if it < warmup_iters:
        return learning_rate * it / warmup_iters
    # 2) if it > lr_decay_iters, return min learning rate
    if it > lr_decay_iters:
        return min_lr
    # 3) in between, use cosine decay down to min learning rate
    decay_ratio = (it - warmup_iters) / (lr_decay_iters - warmup_iters)
    assert 0 <= decay_ratio <= 1
    coeff = 0.5 * (1.0 + math.cos(math.pi * decay_ratio)) # coeff ranges 0..1
    return min_lr + coeff * (learning_rate - min_lr)

# Logging setup
if wandb_log and master_process:
    import wandb
    wandb.init(project=wandb_project, name=wandb_run_name, config=config)

# Training loop
local_iter_num = 0  # Number of iterations in the lifetime of this process
raw_model = model.module if ddp else model  # Unwrap DDP container if needed

t0 = time.time()
while True:
    if iter_num % train_file_update_interval == 0:
        current_train_file_index = random.randint(0, len(train_files) - 1)
        print(f"Switched to file: {train_files[current_train_file_index]}")

    # Determine and set the learning rate for this iteration
    lr = get_lr(iter_num) if decay_lr else learning_rate
    for param_group in optimizer.param_groups:
        param_group['lr'] = lr

    # Evaluate the loss on train/val sets and write checkpoints
    if iter_num % eval_interval == 0 and master_process:
        losses = estimate_loss()
        print(f"\nstep {iter_num}: 'val' loss {losses['val']:.4f}") # Temporary hack
        #print(f"step {iter_num}: train loss {losses['train']:.4f}, val loss {losses['val']:.4f}")
        if wandb_log:
            wandb.log({
                "iter": iter_num,
                #"train/loss": losses['train'], # Temporary hack
                "val/loss": losses['val'],
                "lr": lr,
            })
        if losses['val'] < best_val_loss or always_save_checkpoint:
            best_val_loss = losses['val']
            if iter_num > 0:
                checkpoint = {
                    'model': raw_model.state_dict(),
                    'optimizer': optimizer.state_dict(),
                    'model_args': mamba_config,
                    'iter_num': iter_num,
                    'best_val_loss': best_val_loss,
                    'config': config,
                }
                print(f"saving checkpoint to {out_dir}\n")
                torch.save(checkpoint, os.path.join(out_dir, 'ckpt.pt'))
                
    if iter_num == 0 and eval_only:
        break

    # Forward and backward pass
    for micro_step in range(gradient_accumulation_steps):
        if ddp:
            model.require_backward_grad_sync = (micro_step == gradient_accumulation_steps - 1)

        sequences = get_batch('train')  # Fetch the training data
        with ctx:
            logits = model(sequences[:, :-1])  # Forward pass, exclude last token for input
            # Compute loss (assuming next token prediction task)
            targets = sequences[:, 1:].reshape(-1)  # Shifted by one for next token prediction
            loss = F.cross_entropy(logits.view(-1, logits.size(-1)), targets)
            loss = loss / gradient_accumulation_steps

        scaler.scale(loss).backward()

    # clip the gradient
    if grad_clip != 0.0:
        scaler.unscale_(optimizer)
        torch.nn.utils.clip_grad_norm_(model.parameters(), grad_clip)
    # step the optimizer and scaler if training in fp16
    scaler.step(optimizer)
    scaler.update()
    # flush the gradients as soon as we can, no need for this memory anymore
    optimizer.zero_grad(set_to_none=True)
    
    # timing and logging
    t1 = time.time()
    dt = t1 - t0
    t0 = t1
    if iter_num % log_interval == 0 and master_process:
        # get loss as float. note: this is a CPU-GPU sync point
        # scale up to undo the division above, approximating the true total loss (exact would have been a sum)
        lossf = loss.item() * gradient_accumulation_steps
        print(f"iter {iter_num}: loss {lossf:.4f}, time {dt*1000:.2f}ms")
        if wandb_log:
            wandb.log({
                "iter": iter_num,
                "train/loss": lossf,
                "lr": lr,
            })
    iter_num += 1
    local_iter_num += 1

    # termination conditions
    if iter_num > max_iters:
        checkpoint = {
            'model': raw_model.state_dict(),
            'optimizer': optimizer.state_dict(),
            'model_args': mamba_config,
            'iter_num': iter_num,
            'best_val_loss': best_val_loss,
            'config': config,
        }
        print(f"saving checkpoint to {out_dir}")
        torch.save(checkpoint, os.path.join(out_dir, 'ckpt_final.pt'))
        break

    
    
if ddp:
    destroy_process_group()