[autoparallel] gpt2 autoparallel examples (#2267)

* [autoparallel] gpt2 autoparallel examples

* polish code

* polish code

colossalai/auto_parallel/tensor_shard/initialize.py

@@ -172,7 +172,8 @@ def initialize_model(model: nn.Module,
                      memory_budget: float = -1.0,
                      save_solver_solution: bool = False,
                      load_solver_solution: bool = False,
-                     solution_path: str = None):
+                     solution_path: str = None,
+                     return_solution: bool = False):
     '''
     This method is used to initialize the sharded model which could be used as normal pytorch model.
 
@@ -187,6 +188,9 @@ def initialize_model(model: nn.Module,
         load_solver_solution(optional): if the load_solver_solution is True, the solution will be loaded
             from the solution_path.
         solution_path(optional): the path to save or load the solution.
+        return_solution(optional): if the return_solution is True, the solution will be returned. The returned
+            solution will be used to debug or help to analyze the sharding result. Therefore, we will not just
+            return a series of integers, but return the best strategies.
     '''
     tracer = ColoTracer()
 
@@ -204,7 +208,14 @@ def initialize_model(model: nn.Module,
     gm, sharding_spec_dicts = transform_to_sharded_model(gm, solution, device_mesh, strategies_constructor)
     model_to_return = ModuleWrapper(gm, *sharding_spec_dicts)
 
-    return model_to_return
+    if return_solution:
+        solution_to_return = []
+        nodes = [strategies_vector.node for strategies_vector in strategies_constructor.leaf_strategies]
+        for index, node in enumerate(nodes):
+            solution_to_return.append(f'{node.name} {node.strategies_vector[solution[index]].name}')
+        return model_to_return, solution_to_return
+    else:
+        return model_to_return
 
 
 def autoparallelize(model: nn.Module,
@@ -216,6 +227,7 @@ def autoparallelize(model: nn.Module,
                     save_solver_solution: bool = False,
                     load_solver_solution: bool = False,
                     solver_solution_path: str = None,
+                    return_solution: bool = False,
                     memory_budget: float = -1.0):
     '''
     This method is used to initialize the device mesh, extract the meta_args, and
@@ -238,18 +250,26 @@ def autoparallelize(model: nn.Module,
         load_solver_solution(optional): if the load_solver_solution is True, the solution will be loaded
             from the solution_path.
         solver_solution_path(optional): the path to save or load the solution.
+        return_solution(optional): if the return_solution is True, the solution will be returned.
         memory_budget(optional): the max cuda memory could be used. If the memory budget is -1.0,
             the memory budget will be infinity.
     '''
     device_mesh = initialize_device_mesh(alpha_beta_dict=alpha_beta_dict, logical_mesh_shape=logical_mesh_shape)
     if meta_args is None:
         meta_args = extract_meta_args_from_dataloader(data_loader, data_process_func)
-    model = initialize_model(model,
+
-                             meta_args,
+    rst_to_unpack = initialize_model(model,
-                             device_mesh,
+                                     meta_args,
-                             save_solver_solution=save_solver_solution,
+                                     device_mesh,
-                             load_solver_solution=load_solver_solution,
+                                     save_solver_solution=save_solver_solution,
-                             solver_solution_path=solver_solution_path,
+                                     load_solver_solution=load_solver_solution,
-                             memory_budget=memory_budget)
+                                     solver_solution_path=solver_solution_path,
-
+                                     return_solution=return_solution,
-    return model
+                                     memory_budget=memory_budget)
+
+    if return_solution:
+        model, solution = rst_to_unpack
+        return model, solution
+    else:
+        model = rst_to_unpack
+        return model

examples/language/gpt/auto_parallel_with_gpt/README.md

@@ -0,0 +1,44 @@
+# Auto-Parallelism with GPT2
+
+## Requirements
+
+Before you can launch training, you need to install the following requirements.
+
+### Install PyTorch
+
+```bash
+#conda
+conda install pytorch==1.12.0 torchvision==0.13.0 torchaudio==0.12.0 cudatoolkit=11.3 -c pytorch
+#pip
+pip install torch==1.12.0+cu113 torchvision==0.13.0+cu113 torchaudio==0.12.0 --extra-index-url https://download.pytorch.org/whl/cu113
+```
+
+### Install [Colossal-AI v0.1.12](https://colossalai.org/download/) From Official Website
+
+```bash
+pip install colossalai==0.1.12+torch1.12cu11.3 -f https://release.colossalai.org
+```
+
+### Install transformers
+
+```bash
+pip install transformers
+```
+
+### Install pulp and coin-or-cbc
+
+```bash
+pip install pulp
+conda install -c conda-forge coin-or-cbc
+```
+
+## Dataset
+
+For simplicity, the input data is randonly generated here.
+
+## Training
+
+```bash
+#Run the auto parallel resnet example with 4 GPUs with a dummy dataset.
+colossalai run --nproc_per_node 4 auto_parallel_with_gpt.py
+```

examples/language/gpt/auto_parallel_with_gpt/auto_parallel_with_gpt.py

@@ -0,0 +1,109 @@
+from functools import partial
+from time import time
+from typing import Dict, Optional, Tuple, Union
+
+import psutil
+import torch
+import torch.multiprocessing as mp
+import torch.nn as nn
+import transformers
+from gpt_modules import GPT2LMHeadModel, GPTLMLoss
+from torch.fx import GraphModule
+
+from colossalai.auto_parallel.tensor_shard.initialize import autoparallelize, initialize_model
+from colossalai.core import global_context as gpc
+from colossalai.device.device_mesh import DeviceMesh
+from colossalai.initialize import launch_from_torch
+from colossalai.logging import disable_existing_loggers, get_dist_logger
+
+BATCH_SIZE = 8
+SEQ_LENGTH = 128
+HIDDEN_DIM = 3072
+NUM_HEADS = 16
+NUM_LAYERS = 1
+VOCAB_SIZE = 50257
+NUM_STEPS = 10
+FP16 = False
+
+
+def get_cpu_mem():
+    return psutil.Process().memory_info().rss / 1024**2
+
+
+def get_gpu_mem():
+    return torch.cuda.memory_allocated() / 1024**2
+
+
+def get_mem_info(prefix=''):
+    return f'{prefix}GPU memory usage: {get_gpu_mem():.2f} MB, CPU memory usage: {get_cpu_mem():.2f} MB'
+
+
+def get_tflops(model_numel, batch_size, seq_len, step_time):
+    # Tflops_per_GPU = global_batch * global_numel * seq_len * 8 / #gpu
+    return model_numel * batch_size * seq_len * 8 / 1e12 / (step_time + 1e-12) / 4
+
+
+# Randomly Generated Data
+def get_data(batch_size, seq_len, vocab_size):
+    input_ids = torch.randint(0, vocab_size, (batch_size, seq_len), device=torch.cuda.current_device())
+    attention_mask = torch.ones_like(input_ids)
+    return input_ids, attention_mask
+
+
+def main():
+    disable_existing_loggers()
+    launch_from_torch(config={})
+    logger = get_dist_logger()
+    config = transformers.GPT2Config(n_position=SEQ_LENGTH, n_layer=NUM_LAYERS, n_head=NUM_HEADS, n_embd=HIDDEN_DIM)
+    if FP16:
+        model = GPT2LMHeadModel(config=config).half().to('cuda')
+    else:
+        model = GPT2LMHeadModel(config=config).to('cuda')
+    global_numel = sum([p.numel() for p in model.parameters()])
+
+    meta_input_sample = {
+        'input_ids': torch.zeros((BATCH_SIZE, SEQ_LENGTH), dtype=torch.int64).to('meta'),
+        'attention_mask': torch.zeros((BATCH_SIZE, SEQ_LENGTH), dtype=torch.int64).to('meta'),
+    }
+
+    # Both device mesh initialization and model initialization will be integrated into autoparallelize
+    physical_mesh_id = torch.arange(0, 4)
+    mesh_shape = (2, 2)
+    device_mesh = DeviceMesh(physical_mesh_id, mesh_shape, init_process_group=True)
+
+    # Enable auto-parallel
+    gm, solution = initialize_model(model, meta_input_sample, device_mesh, return_solution=True)
+
+    # print solution on rank 0
+    if gpc.get_global_rank() == 0:
+        for node_strategy in solution:
+            print(node_strategy)
+
+    # build criterion
+    criterion = GPTLMLoss()
+
+    optimizer = torch.optim.Adam(gm.parameters(), lr=0.01)
+    logger.info(get_mem_info(prefix='After init model, '), ranks=[0])
+    get_tflops_func = partial(get_tflops, global_numel, BATCH_SIZE, SEQ_LENGTH)
+    torch.cuda.synchronize()
+    model.train()
+
+    for n in range(10):
+        # we just use randomly generated data here
+        input_ids, attn_mask = get_data(BATCH_SIZE, SEQ_LENGTH, VOCAB_SIZE)
+        optimizer.zero_grad()
+        start = time()
+        outputs = gm(input_ids, attn_mask)
+        loss = criterion(outputs, input_ids)
+        loss.backward()
+        optimizer.step()
+        torch.cuda.synchronize()
+        step_time = time() - start
+        logger.info(
+            f'[{n+1}/{NUM_STEPS}] Loss:{loss.item():.3f}, Step time: {step_time:.3f}s, TFLOPS: {get_tflops_func(step_time):.3f}',
+            ranks=[0])
+    torch.cuda.synchronize()
+
+
+if __name__ == '__main__':
+    main()

examples/language/gpt/auto_parallel_with_gpt/gpt_modules.py

@@ -0,0 +1,253 @@
+from typing import Optional, Tuple, Union
+
+import torch
+import torch.nn as nn
+from transformers.activations import ACT2FN
+from transformers.models.gpt2.modeling_gpt2 import BaseModelOutputWithPastAndCrossAttentions, GPT2PreTrainedModel
+from transformers.pytorch_utils import Conv1D
+
+
+class GPT2MLP(nn.Module):
+
+    def __init__(self, intermediate_size, config):
+        super().__init__()
+        embed_dim = config.hidden_size
+        self.c_fc = Conv1D(intermediate_size, embed_dim)
+        self.c_proj = Conv1D(embed_dim, intermediate_size)
+        self.act = ACT2FN[config.activation_function]
+        self.dropout = nn.Dropout(config.resid_pdrop)
+
+    def forward(self, hidden_states: Optional[Tuple[torch.FloatTensor]]) -> torch.FloatTensor:
+        hidden_states = self.c_fc(hidden_states)
+        hidden_states = self.act(hidden_states)
+        hidden_states = self.c_proj(hidden_states)
+        return hidden_states
+
+
+# The reason Why we don't import GPT2Attention from transformers directly is that:
+# 1. The tracer will not work correctly when we feed meta_args and concrete_args at same time,
+# so we have to build the customized GPT2Attention class and remove the conditional branch manually.
+# 2. The order of split and view op has been changed in the customized GPT2Attention class, the new
+# order is same as megatron-lm gpt model.
+class GPT2Attention(nn.Module):
+
+    def __init__(self, config, layer_idx=None):
+        super().__init__()
+
+        max_positions = config.max_position_embeddings
+        self.register_buffer(
+            "bias",
+            torch.tril(torch.ones((max_positions, max_positions),
+                                  dtype=torch.uint8)).view(1, 1, max_positions, max_positions),
+        )
+        self.register_buffer("masked_bias", torch.tensor(-1e4))
+
+        self.embed_dim = config.hidden_size
+        self.num_heads = config.num_attention_heads
+        self.head_dim = self.embed_dim // self.num_heads
+        self.split_size = self.embed_dim
+        self.scale_attn_weights = config.scale_attn_weights
+
+        # Layer-wise attention scaling, reordering, and upcasting
+        self.scale_attn_by_inverse_layer_idx = config.scale_attn_by_inverse_layer_idx
+        self.layer_idx = layer_idx
+
+        self.c_attn = Conv1D(3 * self.embed_dim, self.embed_dim)
+        self.c_proj = Conv1D(self.embed_dim, self.embed_dim)
+
+        self.attn_dropout = nn.Dropout(config.attn_pdrop)
+        self.resid_dropout = nn.Dropout(config.resid_pdrop)
+
+        self.pruned_heads = set()
+
+    def _attn(self, query, key, value, attention_mask=None, head_mask=None):
+        attn_weights = torch.matmul(query, key.transpose(-1, -2))
+
+        if self.scale_attn_weights:
+            attn_weights = attn_weights / (value.size(-1)**0.5)
+
+        # Layer-wise attention scaling
+        if self.scale_attn_by_inverse_layer_idx:
+            attn_weights = attn_weights / float(self.layer_idx + 1)
+
+        # if only "normal" attention layer implements causal mask
+        query_length, key_length = query.size(-2), key.size(-2)
+        causal_mask = self.bias[:, :, key_length - query_length:key_length, :key_length].to(torch.bool)
+        attn_weights = torch.where(causal_mask, attn_weights, self.masked_bias.to(attn_weights.dtype))
+
+        if attention_mask is not None:
+            # Apply the attention mask
+            attn_weights = attn_weights + attention_mask
+
+        attn_weights = nn.functional.softmax(attn_weights, dim=-1)
+        attn_weights = attn_weights.type(value.dtype)
+
+        # Mask heads if we want to
+        if head_mask is not None:
+            attn_weights = attn_weights * head_mask
+
+        attn_output = torch.matmul(attn_weights, value)
+
+        return attn_output, attn_weights
+
+    def _split_heads(self, tensor, num_heads, attn_head_size):
+        new_shape = tensor.size()[:-1] + (num_heads, attn_head_size)
+        tensor = tensor.view(new_shape)
+        return tensor.permute(0, 2, 1, 3)    # (batch, head, seq_length, head_features)
+
+    def _merge_heads(self, tensor, num_heads, attn_head_size):
+        tensor = tensor.permute(0, 2, 1, 3).contiguous()
+        new_shape = tensor.size()[:-2] + (num_heads * attn_head_size,)
+        return tensor.view(new_shape)
+
+    def forward(
+        self,
+        hidden_states: Optional[Tuple[torch.FloatTensor]],
+        attention_mask: Optional[torch.FloatTensor] = None,
+        head_mask: Optional[torch.FloatTensor] = None,
+    ) -> Tuple[Union[torch.Tensor, Tuple[torch.Tensor]], ...]:
+
+        qkv = self.c_attn(hidden_states)
+        query, key, value = self._split_heads(qkv, self.num_heads, 3 * self.head_dim).split(self.head_dim, dim=3)
+        present = (key, value)
+        attn_output, attn_weights = self._attn(query, key, value, attention_mask, head_mask)
+        attn_output = self._merge_heads(attn_output, self.num_heads, self.head_dim)
+        attn_output = self.c_proj(attn_output)
+        return attn_output
+
+
+class GPT2Block(nn.Module):
+
+    def __init__(self, config, layer_idx=None):
+        super().__init__()
+        hidden_size = config.hidden_size
+        inner_dim = config.n_inner if config.n_inner is not None else 4 * hidden_size
+        self.ln_1 = nn.LayerNorm(hidden_size, eps=config.layer_norm_epsilon)
+        self.attn = GPT2Attention(config, layer_idx=layer_idx)
+        self.ln_2 = nn.LayerNorm(hidden_size, eps=config.layer_norm_epsilon)
+        self.mlp = GPT2MLP(inner_dim, config)
+
+    def forward(
+        self,
+        hidden_states: Optional[Tuple[torch.FloatTensor]],
+        attention_mask: Optional[torch.FloatTensor] = None,
+        head_mask: Optional[torch.FloatTensor] = None,
+    ) -> Union[Tuple[torch.Tensor], Optional[Tuple[torch.Tensor, Tuple[torch.FloatTensor, ...]]]]:
+        residual = hidden_states
+        hidden_states = self.ln_1(hidden_states)
+        attn_outputs = self.attn(
+            hidden_states,
+            attention_mask=attention_mask,
+            head_mask=head_mask,
+        )
+        # residual connection
+        hidden_states = attn_outputs + residual
+        residual = hidden_states
+        hidden_states = self.ln_2(hidden_states)
+        feed_forward_hidden_states = self.mlp(hidden_states)
+        # residual connection
+        hidden_states = residual + feed_forward_hidden_states
+
+        return hidden_states
+
+
+class GPT2Model(GPT2PreTrainedModel):
+
+    def __init__(self, config):
+        super().__init__(config)
+
+        self.embed_dim = config.hidden_size
+
+        self.wte = nn.Embedding(config.vocab_size, self.embed_dim)
+        self.wpe = nn.Embedding(config.max_position_embeddings, self.embed_dim)
+
+        self.drop = nn.Dropout(config.embd_pdrop)
+        self.h = nn.ModuleList([GPT2Block(config, layer_idx=i) for i in range(config.num_hidden_layers)])
+        self.ln_f = nn.LayerNorm(self.embed_dim, eps=config.layer_norm_epsilon)
+
+        # Initialize weights and apply final processing
+        self.post_init()
+
+    def forward(
+        self,
+        input_ids: Optional[torch.LongTensor] = None,
+        attention_mask: Optional[torch.FloatTensor] = None,
+        head_mask: Optional[torch.FloatTensor] = None,
+    ) -> Union[Tuple, BaseModelOutputWithPastAndCrossAttentions]:
+        input_shape = input_ids.size()
+        input_ids = input_ids.view(-1, input_shape[-1])
+        batch_size = input_ids.shape[0]
+
+        device = input_ids.device
+
+        past_length = 0
+        past_key_values = tuple([None] * len(self.h))
+
+        position_ids = torch.arange(past_length, input_shape[-1] + past_length, dtype=torch.long, device=device)
+        position_ids = position_ids.unsqueeze(0).view(-1, input_shape[-1])
+
+        # GPT2Attention mask.
+        attention_mask = attention_mask.view(batch_size, -1)
+        attention_mask = attention_mask[:, None, None, :]
+        attention_mask = attention_mask.to(dtype=self.dtype)    # fp16 compatibility
+        attention_mask = (1.0 - attention_mask) * -10000.0
+
+        encoder_attention_mask = None
+
+        # Prepare head mask if needed
+        # 1.0 in head_mask indicate we keep the head
+        # attention_probs has shape bsz x n_heads x N x N
+        # head_mask has shape n_layer x batch x n_heads x N x N
+        head_mask = self.get_head_mask(head_mask, self.config.n_layer)
+        inputs_embeds = self.wte(input_ids)
+        position_embeds = self.wpe(position_ids)
+
+        hidden_states = inputs_embeds + position_embeds
+
+        output_shape = input_shape + (hidden_states.size(-1),)
+
+        for i, (block, layer_past) in enumerate(zip(self.h, past_key_values)):
+            outputs = block(hidden_states, attention_mask=attention_mask, head_mask=head_mask[i])
+            hidden_states = outputs
+
+        hidden_states = self.ln_f(hidden_states)
+        hidden_states = hidden_states.view(output_shape)
+
+        return hidden_states
+
+
+class GPT2LMHeadModel(GPT2PreTrainedModel):
+
+    def __init__(self, config):
+        super().__init__(config)
+        self.transformer = GPT2Model(config)
+        self.lm_head = nn.Linear(config.n_embd, config.vocab_size, bias=False)
+
+        # Initialize weights and apply final processing
+        self.post_init()
+
+    def forward(
+        self,
+        input_ids: Optional[torch.LongTensor] = None,
+        attention_mask: Optional[torch.FloatTensor] = None,
+    ):
+        transformer_outputs = self.transformer(
+            input_ids=input_ids,
+            attention_mask=attention_mask,
+        )
+        lm_logits = self.lm_head(transformer_outputs)
+
+        return lm_logits
+
+
+class GPTLMLoss(nn.Module):
+
+    def __init__(self):
+        super().__init__()
+        self.loss_fn = nn.CrossEntropyLoss()
+
+    def forward(self, logits, labels):
+        shift_logits = logits[..., :-1, :].contiguous()
+        shift_labels = labels[..., 1:].contiguous()
+        # Flatten the tokens
+        return self.loss_fn(shift_logits.view(-1, shift_logits.size(-1)), shift_labels.view(-1))

examples/language/gpt/auto_parallel_with_gpt/requirements.txt

@@ -0,0 +1,4 @@
+colossalai >= 0.1.12
+torch >= 1.8.1
+transformers >= 4.231
+PuLP >= 2.7.0