improvements

RWKV-v4neo/src/dataset.py

@@ -2,12 +2,13 @@
 # The RWKV Language Model - https://github.com/BlinkDL/RWKV-LM
 ########################################################################################################
 
-import json
+import json, math
 import numpy as np
 import torch
 from torch.utils.data import Dataset
 from pytorch_lightning.utilities import rank_zero_info
 from .binidx import MMapIndexedDataset
+from .utils import MaybeIsPrime
 
 
 class MyDataset(Dataset):
@@ -20,6 +21,18 @@ class MyDataset(Dataset):
             print("current vocab size =", self.vocab_size, "(make sure it's correct)")
             self.data_size = len(self.data._bin_buffer) // 2
             print(f"data has {self.data_size} tokens.")
+
+            if args.my_pile_mode > 0:
+                assert self.data_size == 332115325534 and self.vocab_size == 50277 and args.ctx_len == 1024
+                self.samples_per_epoch = args.epoch_steps * int(args.devices) * args.micro_bsz
+                assert self.samples_per_epoch == 40320
+                print("########## Pile 20b-tokenized mode {args.my_pile_mode} ##########")
+                self.magic_prime = 324331313
+                dataset_slot = self.data_size // args.ctx_len
+                assert MaybeIsPrime(self.magic_prime)
+                assert self.magic_prime % 3 == 2
+                assert self.magic_prime / dataset_slot > 0.999999 and self.magic_prime / dataset_slot <= 1
+
         elif args.data_type == "numpy":
             self.data = np.load(args.data_file).astype("int")
             self.vocab_size = args.vocab_size
@@ -48,15 +61,29 @@ class MyDataset(Dataset):
             self.itos = {i: ch for i, ch in enumerate(unique)}
 
     def __len__(self):
-        return self.args.epoch_steps * int(self.args.devices) * self.args.micro_bsz
+        return self.args.epoch_steps * self.args.micro_bsz
 
     def __getitem__(self, idx):
         #
         # we are cheating: pick a random spot in dataset
         #
+        rank = self.global_rank
+        epoch = self.real_epoch
+        world_size = self.world_size
+        # print(f"epoch {epoch} idx {idx} rank {rank}/{world_size}")
+
         ctx_len = self.args.ctx_len
         req_len = ctx_len + 1
-        i = np.random.randint(0, self.data_size - req_len)
+
+        if self.args.my_pile_mode > 0:
+            ii = 1 + epoch * self.samples_per_epoch + (idx * world_size) + rank
+            factor = (math.sqrt(5) - 1) / 2
+            factor = int(self.magic_prime * factor)
+            i = ((factor * ii * ii * ii) % self.magic_prime) * ctx_len
+            # print(f"epoch {epoch} idx {idx} rank {rank}/{world_size} ii {ii} pos {round(i / self.data_size, 3)}")
+        else:
+            i = np.random.randint(0, self.data_size - req_len)
+
         if "MMapIndexedDataset" in str(type(self.data)):
             dix = self.data.get(idx=0, offset=i, length=req_len).astype(int)
         elif "numpy" in str(type(self.data)):

RWKV-v4neo/src/model.py

@@ -2,7 +2,7 @@
 # The RWKV Language Model - https://github.com/BlinkDL/RWKV-LM
 ########################################################################################################
 
-import os, math, gc, time
+import os, math, gc
 from re import L
 import torch
 import torch.nn as nn
@@ -20,7 +20,7 @@ def __nop(ob):
 
 MyModule = nn.Module
 MyFunction = __nop
-if os.environ["RWKV_JIT"] == "1":
+if os.environ["RWKV_JIT_ON"] == "1":
     MyModule = torch.jit.ScriptModule
     MyFunction = torch.jit.script_method
 
@@ -273,9 +273,31 @@ class RWKV(pl.LightningModule):
             self.register_buffer("copy_mask", torch.tril(torch.ones(args.ctx_len, args.ctx_len)))
 
     def configure_optimizers(self):
+        lr_1x = set()
+        lr_2x = set()
+        lr_3x = set()
+        for n, p in self.named_parameters():
+            if ("time_mix" in n) and (self.args.my_pile_mode == 2):
+                lr_2x.add(n)
+            elif "time_decay" in n:
+                lr_2x.add(n)
+            elif "time_first" in n:
+                lr_3x.add(n)
+            else:
+                lr_1x.add(n)
+        lr_1x = sorted(list(lr_1x))
+        lr_2x = sorted(list(lr_2x))
+        lr_3x = sorted(list(lr_3x))
+        # print('1x', lr_1x)
+        # print('2x', lr_2x)
+        # print('3x', lr_3x)
+        param_dict = {n: p for n, p in self.named_parameters()}
         optim_groups = [
-            {"params": [p for n, p in self.named_parameters()], "weight_decay": 0.0},
+            {"params": [param_dict[n] for n in lr_1x], "weight_decay": 0.0, "my_lr_scale": 1.0},
+            {"params": [param_dict[n] for n in lr_2x], "weight_decay": 0.0, "my_lr_scale": 2.0},
+            {"params": [param_dict[n] for n in lr_3x], "weight_decay": 0.0, "my_lr_scale": 3.0},
         ]
+
         if self.deepspeed_offload:
             return DeepSpeedCPUAdam(optim_groups, lr=self.args.lr_init, betas=self.args.betas, eps=self.args.adam_eps, bias_correction=True, adamw_mode=False, weight_decay=0, amsgrad=False)
         return FusedAdam(optim_groups, lr=self.args.lr_init, betas=self.args.betas, eps=self.args.adam_eps, bias_correction=True, adam_w_mode=False, weight_decay=0, amsgrad=False)
@@ -326,25 +348,13 @@ class RWKV(pl.LightningModule):
         idx, targets = batch
         logits = self(idx)
         loss = F.cross_entropy(logits.view(-1, logits.size(-1)), targets.view(-1))
-
-        if self.trainer.global_rank == 0:
-            t_now = time.time_ns()
-            try:
-                t_cost = (t_now - self.trainer.my_time_ns) / 1e9
-                self.log("REAL it/s", 1.0 / t_cost, prog_bar=True, on_step=True)
-                self.log("token/s", args.ctx_len * float(args.devices) * args.micro_bsz / t_cost, prog_bar=True, on_step=True)
-            except:
-                pass
-            self.trainer.my_time_ns = t_now
-            self.trainer.my_loss = loss.item()
-            self.trainer.my_loss_sum += self.trainer.my_loss
-            self.trainer.my_loss_count += 1
-            self.trainer.my_epoch_loss = self.trainer.my_loss_sum / self.trainer.my_loss_count
-            self.log("lr", self.trainer.my_lr, prog_bar=True, on_step=True)
-            self.log("loss", self.trainer.my_epoch_loss, prog_bar=True, on_step=True)
-
         return L2Wrap.apply(loss, logits)
 
+    def training_step_end(self, batch_parts):
+        all = self.all_gather(batch_parts)
+        if self.trainer.is_global_zero:
+            self.trainer.my_loss_all = all
+
     def generate_init_weight(self):
         print(
             f"""

RWKV-v4neo/src/utils.py

@@ -0,0 +1,50 @@
+import random
+
+
+def MaybeIsPrime(number):
+    if FermatPrimalityTest(number) and MillerRabinPrimalityTest(number):
+        return True
+    else:
+        return False
+
+
+def FermatPrimalityTest(number):
+    if number > 1:
+        for time in range(3):
+            randomNumber = random.randint(2, number) - 1
+            if pow(randomNumber, number - 1, number) != 1:
+                return False
+        return True
+    else:
+        return False
+
+
+def MillerRabinPrimalityTest(number):
+    if number == 2:
+        return True
+    elif number == 1 or number % 2 == 0:
+        return False
+    oddPartOfNumber = number - 1
+    timesTwoDividNumber = 0
+    while oddPartOfNumber % 2 == 0:
+        oddPartOfNumber = oddPartOfNumber // 2
+        timesTwoDividNumber = timesTwoDividNumber + 1
+
+    for time in range(3):
+        while True:
+            randomNumber = random.randint(2, number) - 1
+            if randomNumber != 0 and randomNumber != 1:
+                break
+
+        randomNumberWithPower = pow(randomNumber, oddPartOfNumber, number)
+
+        if (randomNumberWithPower != 1) and (randomNumberWithPower != number - 1):
+            iterationNumber = 1
+
+            while (iterationNumber <= timesTwoDividNumber - 1) and (randomNumberWithPower != number - 1):
+                randomNumberWithPower = pow(randomNumberWithPower, 2, number)
+                iterationNumber = iterationNumber + 1
+            if randomNumberWithPower != (number - 1):
+                return False
+
+    return True

RWKV-v4neo/train.py

@@ -4,10 +4,11 @@
 
 if __name__ == "__main__":
     print("\n!!! NOTE: THIS IS STILL WIP !!!\n")
-    import os, warnings, math, datetime, sys
+    import os, warnings, math, datetime, sys, time
     import numpy as np
     from argparse import ArgumentParser
     import torch
+    from torch.utils.data import DataLoader
     import deepspeed
     import pytorch_lightning as pl
     from pytorch_lightning import Trainer
@@ -83,10 +84,12 @@ if __name__ == "__main__":
     parser.add_argument("--adam_eps", default=1e-8, type=float)
 
     parser.add_argument("--grad_cp", default=0, type=int)  # gradient checkpt: saves VRAM, but slower
+    parser.add_argument("--my_pile_mode", default=0, type=int)  # my special pile mode
 
     args = parser.parse_args()
     args.my_timestamp = datetime.datetime.today().strftime("%Y-%m-%d-%H-%M-%S")
     args.enable_checkpointing = False
+    args.replace_sampler_ddp = False
     args.logger = False
     args.gradient_clip_val = 1.0
     args.num_sanity_val_steps = 0
@@ -95,6 +98,12 @@ if __name__ == "__main__":
     args.max_epochs = -1  # continue forever
     args.betas = (args.beta1, args.beta2)
 
+    if args.my_pile_mode > 0:
+        args.epoch_steps = 40320 // (int(args.devices) * args.micro_bsz)
+        assert args.epoch_steps * int(args.devices) * args.micro_bsz == 40320
+        if args.my_pile_mode == 2:
+            assert args.lr_final == args.lr_init
+
     samples_per_epoch = args.epoch_steps * int(args.devices) * args.micro_bsz
     tokens_per_epoch = samples_per_epoch * args.ctx_len
     rank_zero_info(
@@ -138,9 +147,9 @@ if __name__ == "__main__":
     if args.precision == "fp16":
         rank_zero_info("\n\nNote: you are using fp16 (might overflow). Try bf16 / tf32 for stable training.\n\n")
 
-    os.environ["RWKV_JIT"] = "1"
+    os.environ["RWKV_JIT_ON"] = "1"
     if "deepspeed_stage_3" in args.strategy:
-        os.environ["RWKV_JIT"] = "0"
+        os.environ["RWKV_JIT_ON"] = "0"
 
     import torch
 
@@ -170,9 +179,37 @@ if __name__ == "__main__":
             args = self.args
             g_step = trainer.global_step
 
-            # logging
-            if trainer.global_rank == 0:
-                if g_step == 0:
+            # LR schedule
+            w_step = args.warmup_steps
+            if g_step < w_step:
+                lr = args.lr_init * (g_step / w_step)
+            else:
+                if args.lr_final == args.lr_init:
+                    lr = args.lr_init
+                else:
+                    progress = (g_step - w_step) / (args.epoch_count * args.epoch_steps - w_step - 1)
+                    progress = min(1, max(0, progress))
+
+                    if args.lr_final == 0 or args.lr_init == 0:  # linear decay
+                        lr = args.lr_init + (args.lr_final - args.lr_init) * progress
+                    else:  # exp decay
+                        lr = args.lr_init * math.exp(math.log(args.lr_final / args.lr_init) * pow(progress, 1))
+
+            for param_group in trainer.optimizers[0].param_groups:
+                if self.args.my_pile_mode == 0:
+                    param_group["lr"] = lr * param_group["my_lr_scale"]
+                elif self.args.my_pile_mode == 2:
+                    if param_group["my_lr_scale"] > 1:
+                        param_group["lr"] = lr * 5
+                    else:
+                        param_group["lr"] = lr
+                # print(param_group["lr"], param_group["my_lr_scale"])
+
+            trainer.my_lr = lr
+            # rank_zero_info(f"{g_step} {lr}")
+
+            if g_step == 0:
+                if trainer.is_global_zero:  # logging
                     trainer.my_loss_sum = 0
                     trainer.my_loss_count = 0
                     trainer.my_log = open(args.proj_dir + "/train_log.txt", "a")
@@ -196,39 +233,42 @@ if __name__ == "__main__":
                         )
                         trainer.my_wandb = wandb
 
-            # LR schedule
-            w_step = args.warmup_steps
-            if g_step < w_step:
-                lr = args.lr_init * (g_step / w_step)
-            else:
-                progress = (g_step - w_step) / (args.epoch_count * args.epoch_steps - w_step - 1)
-                progress = min(1, max(0, progress))
-
-                if args.lr_final == 0 or args.lr_init == 0:  # linear decay
-                    lr = args.lr_init + (args.lr_final - args.lr_init) * progress
-                else:  # exp decay
-                    lr = args.lr_init * math.exp(math.log(args.lr_final / args.lr_init) * pow(progress, 1))
-
-            for param_group in trainer.optimizers[0].param_groups:
-                param_group["lr"] = lr
-
-            trainer.my_lr = lr
-            # rank_zero_info(f"{g_step} {lr}")
-
         def on_train_batch_end(self, trainer, pl_module, outputs, batch, batch_idx):
             args = self.args
-            # logging
-            if trainer.global_rank == 0:
+            if trainer.is_global_zero:  # logging
+                t_now = time.time_ns()
+                try:
+                    t_cost = (t_now - trainer.my_time_ns) / 1e9
+                    self.log("REAL it/s", 1.0 / t_cost, prog_bar=True, on_step=True)
+                    self.log("token/s", args.ctx_len * float(args.devices) * args.micro_bsz / t_cost, prog_bar=True, on_step=True)
+                except:
+                    pass
+                trainer.my_time_ns = t_now
+                trainer.my_loss = trainer.my_loss_all.float().mean().item()
+                trainer.my_loss_sum += trainer.my_loss
+                trainer.my_loss_count += 1
+                trainer.my_epoch_loss = trainer.my_loss_sum / trainer.my_loss_count
+                self.log("lr", trainer.my_lr, prog_bar=True, on_step=True)
+                self.log("loss", trainer.my_epoch_loss, prog_bar=True, on_step=True)
+
                 if len(args.wandb) > 0:
                     trainer.my_wandb.log(
                         {"loss": trainer.my_loss, "lr": trainer.my_lr},
                         step=trainer.global_step,
                     )
 
+        def on_train_epoch_start(self, trainer, pl_module):
+            args = self.args
+            dataset = trainer.train_dataloader.dataset.datasets
+            assert "MyDataset" in str(dataset)
+            dataset.global_rank = trainer.global_rank
+            dataset.real_epoch = int(args.epoch_begin + trainer.current_epoch)
+            dataset.world_size = trainer.world_size
+
         def on_train_epoch_end(self, trainer, pl_module):
             args = self.args
-            if trainer.global_rank == 0:
-                if trainer.current_epoch % args.epoch_save == 0 or trainer.current_epoch == args.epoch_count - 1:
+            if trainer.is_global_zero:  # logging & save state_dict
+                if (args.epoch_save > 0 and trainer.current_epoch % args.epoch_save == 0) or trainer.current_epoch == args.epoch_count - 1:
                     torch.save(
                         pl_module.state_dict(),
                         f"{args.proj_dir}/rwkv-{args.epoch_begin + trainer.current_epoch}.pth",
@@ -251,7 +291,6 @@ if __name__ == "__main__":
 
     ########################################################################################################
 
-    from torch.utils.data import DataLoader
     from src.dataset import MyDataset
     from src.model import RWKV
 
@@ -261,8 +300,8 @@ if __name__ == "__main__":
     model = RWKV(args)
 
     if len(args.load_model) == 0:
-        args.load_model = f"{args.proj_dir}/rwkv-init.pth"  # init weights to tmp file
-        generate_init_weight(model, args.load_model)
+        args.load_model = f"{args.proj_dir}/rwkv-init.pth"
+        generate_init_weight(model, args.load_model)  # save initial weights to tmp file
 
     print(f"########## Loading {args.load_model}... ##########")
     load_dict = torch.load(args.load_model, map_location="cpu")
@@ -273,5 +312,7 @@ if __name__ == "__main__":
         callbacks=[train_callback(args)],
     )
 
-    train_loader = DataLoader(train_data, batch_size=args.micro_bsz, num_workers=1)
-    trainer.fit(model, train_loader)
+    # must set shuffle=False, persistent_workers=False (because worker is in another thread)
+    data_loader = DataLoader(train_data, shuffle=False, pin_memory=True, batch_size=args.micro_bsz, num_workers=1, persistent_workers=False, drop_last=True)
+
+    trainer.fit(model, data_loader)