Support finishing PP inference once eos_token_id is found (#11336)

python/llm/example/GPU/Pipeline-Parallel-Inference/README.md

@@ -11,6 +11,7 @@ To run this example with IPEX-LLM on Intel GPUs, we have some recommended requir
 - [meta-llama/Meta-Llama-3-8B-Instruct](./run_llama_arc_2_card.sh)
 - [Qwen/Qwen1.5-7B-Chat](./run_qwen1.5_arc_2_card.sh)
 - [Qwen/Qwen1.5-14B-Chat](./run_qwen1.5_arc_2_card.sh)
+- [Qwen/Qwen1.5-32B-Chat](./run_qwen1.5_arc_2_card.sh)
 - [baichuan-inc/Baichuan2-7B-Chat](./run_baichuan2_arc_2_card.sh)
 - [baichuan-inc/Baichuan2-13B-Chat](./run_baichuan2_arc_2_card.sh)
 - [microsoft/Phi-3-mini-4k-instruct](./run_phi3_arc_2_card.sh)
@@ -57,7 +58,7 @@ bash run_llama_arc_2_card.sh
 <details>
   <summary> Show Qwen1.5 example </summary>
 
-#### Run Qwen1.5-7B-Chat / Qwen1.5-14B-Chat on two Intel Arc A770
+#### Run Qwen1.5-7B-Chat / Qwen1.5-14B-Chat / Qwen1.5-32B-Chat on two Intel Arc A770
 
 You could specify `--repo-id-or-model-path` in the test script to be the huggingface repo id for Qwen1.5 to be downloaded, or the path to the huggingface checkpoint folder. Besides, you could change `NUM_GPUS` to the number of GPUs you have on your machine.
 

python/llm/example/GPU/Pipeline-Parallel-Inference/generate.py

@@ -46,6 +46,7 @@
                                                  optimize_model=True,
                                                  trust_remote_code=True,
                                                  use_cache=True,
+                                                 torch_dtype=torch.float16,
                                                  pipeline_parallel_stages=args.gpu_num)
 
     # Load tokenizer

python/llm/example/GPU/Pipeline-Parallel-Inference/run_qwen1.5_arc_2_card.sh

@@ -34,3 +34,7 @@ CCL_ZE_IPC_EXCHANGE=sockets torchrun --standalone --nnodes=1 --nproc-per-node $N
 # # To run Qwen1.5-14B-Chat
 # CCL_ZE_IPC_EXCHANGE=sockets torchrun --standalone --nnodes=1 --nproc-per-node $NUM_GPUS \
 #     generate.py --repo-id-or-model-path 'Qwen/Qwen1.5-14B-Chat' --gpu-num $NUM_GPUS
+
+# # To run Qwen1.5-32B-Chat
+# CCL_ZE_IPC_EXCHANGE=sockets torchrun --standalone --nnodes=1 --nproc-per-node $NUM_GPUS \
+#     generate.py --repo-id-or-model-path 'Qwen/Qwen1.5-32B-Chat' --gpu-num $NUM_GPUS

python/llm/src/ipex_llm/transformers/pipeline_parallel.py

@@ -25,6 +25,9 @@
 import numpy as np
 from typing import Callable, List, Optional
 from transformers import GenerationConfig, LogitsProcessorList, StoppingCriteriaList
+from ipex_llm.utils.common import invalidInputError
+import logging
+logger = logging.getLogger(__name__)
 
 # patch GenerationMixin.generate
 from transformers import GenerationMixin
@@ -118,12 +121,34 @@ def generate(
     **kwargs,
 ):
     if hasattr(self, 'pipeline_parallel_stages') and self.pipeline_parallel_stages > 1:
+        # priority: `generation_config` argument > `model.generation_config`
+        if generation_config is None:
+            if (
+                self.generation_config._from_model_config
+                and self.generation_config._original_object_hash == hash(self.generation_config)
+                and self.config._has_non_default_generation_parameters()
+            ):
+                new_generation_config = GenerationConfig.from_model_config(self.config)
+                if new_generation_config != self.generation_config:
+                    self.generation_config = new_generation_config
+            generation_config = self.generation_config
+
+        if generation_config.pad_token_id is None and generation_config.eos_token_id is not None:
+            eos_token_id = generation_config.eos_token_id
+            if isinstance(eos_token_id, list):
+                eos_token_id = eos_token_id[0]
+            logger.warning("Setting `pad_token_id` to `eos_token_id`: "
+                           f"{eos_token_id} for open-end generation.")
+            generation_config.pad_token_id = eos_token_id
+
         if generation_config is not None and generation_config.max_new_tokens is not None:
             max_new_tokens = generation_config.max_new_tokens
         else:
             max_new_tokens = kwargs.get("max_new_tokens", None)
+
         return self.pipeline_parallel_generate(inputs=inputs,
-                                               max_new_tokens=max_new_tokens,)
+                                               max_new_tokens=max_new_tokens,
+                                               generation_config=generation_config,)
 
     return original_generate(self,
                              inputs=inputs,
@@ -143,6 +168,7 @@ def generate(
 def pipeline_parallel_generate(self,
                                inputs: Optional[torch.Tensor] = None,
                                max_new_tokens: int = 32,
+                               generation_config: Optional[GenerationConfig] = None,
                                **kwargs):
     local_rank = dist.get_rank()
     pre_rank = (local_rank - 1) % self.pipeline_parallel_stages
@@ -154,12 +180,22 @@ def pipeline_parallel_generate(self,
     self.first_token_time = 0
     self.next_token_time = []
 
+    pad_token_id = generation_config.pad_token_id
+    eos_token_id = generation_config.eos_token_id
+    if isinstance(eos_token_id, int):
+        eos_token_id = [eos_token_id]
+    eos_token_id_tensor = torch.tensor(eos_token_id).to(inputs.device) \
+        if eos_token_id is not None else None
+
     _input_ids = None
     _past_key_values = None
     bs = inputs.shape[0]
     output_ids = inputs.clone()
 
     step = 0
+    # keep track of which sequences are already finished
+    unfinished_sequences = torch.ones(inputs.shape[0], dtype=torch.long, device=inputs.device)
+    this_peer_finished = False
     while True:
         if step >= max_new_tokens:
             break
@@ -190,6 +226,14 @@ def pipeline_parallel_generate(self,
         _input_ids = next_ids
         output_ids = torch.cat([output_ids, next_ids], dim=-1)
 
+        # finished sentences should have their next token be a padding token
+        next_ids = next_ids.squeeze()
+        if eos_token_id is not None:
+            if pad_token_id is None:
+                invalidInputError(False, "If `eos_token_id` is defined, "
+                                         "make sure that `pad_token_id` is defined.")
+            next_ids = next_ids * unfinished_sequences + pad_token_id * (1 - unfinished_sequences)
+
         if isinstance(outputs.past_key_values, tuple) and local_rank != 0:
             value_placeholder = torch.empty_like((outputs.past_key_values)[-1][0])
             past_key_values_placeholder = tuple(
@@ -204,6 +248,19 @@ def pipeline_parallel_generate(self,
             self.first_token_time = toc - tic
         else:
             self.next_token_time.append(toc - tic)
+
+        # if eos_token was found in one sentence, set sentence to finished
+        if eos_token_id_tensor is not None:
+            unfinished_sequences = unfinished_sequences.mul(
+                next_ids.tile(eos_token_id_tensor.shape[0], 1)
+                .ne(eos_token_id_tensor.unsqueeze(1)).prod(dim=0)
+            )
+            # stop when each sentence is finished
+            if unfinished_sequences.max() == 0:
+                this_peer_finished = True
+        if this_peer_finished:
+            break
+
         step += 1
         if self.device.type == 'xpu':
             torch.xpu.synchronize()