[tnx] fix optimum token selection and sampling (#2233)

engines/python/setup/djl_python/transformers_neuronx_scheduler/optimum_modeling.py

@@ -25,7 +25,7 @@
 from transformers import PretrainedConfig
 from transformers_neuronx import bucket
 from transformers_neuronx.constants import LAYOUT_BSH, LAYOUT_HSB
-from optimum.neuron.generation import TokenSelector
+from djl_python.transformers_neuronx_scheduler.optimum_token_selector import OptimumTokenSelector
 from optimum.neuron.utils.version_utils import check_compiler_compatibility, get_neuronxcc_version
 from optimum.modeling_base import OptimizedModel
 from transformers.generation import StoppingCriteriaList
@@ -238,11 +238,12 @@ def generate(
         self._validate_model_kwargs(model_kwargs)
 
         # Instantiate a TokenSelector for the specified configuration
-        selector = TokenSelector.create(input_ids,
-                                        generation_config,
-                                        self,
-                                        self.max_length,
-                                        stopping_criteria=stopping_criteria)
+        selector = OptimumTokenSelector.create(
+            input_ids,
+            generation_config,
+            self,
+            self.max_length,
+            stopping_criteria=stopping_criteria)
 
         # Verify that the inputs are compatible with the model static input dimensions
         batch_size, sequence_length = input_ids.shape
@@ -280,7 +281,7 @@ def generate(
     def generate_tokens(
         self,
         input_ids: torch.LongTensor,
-        selector: TokenSelector,
+        selector: OptimumTokenSelector,
         batch_size: int,
         attention_mask: Optional[torch.Tensor] = None,
         **model_kwargs,
@@ -291,7 +292,7 @@ def generate_tokens(
         Args:
             input_ids (`torch.LongTensor` of shape `(batch_size, sequence_length)`):
                 The sequence used as a prompt for the generation.
-            selector (`TokenSelector`):
+            selector (`OptimumTokenSelector`):
                 The object implementing the generation logic based on transformers processors and stopping criterias.
             batch_size (`int`):
                 The actual input batch size. Used to avoid generating tokens for padded inputs.

engines/python/setup/djl_python/transformers_neuronx_scheduler/optimum_neuron_scheduler.py

@@ -24,7 +24,6 @@
 from dataclasses import dataclass
 
 from djl_python.transformers_neuronx_scheduler.slot import Slot
-from djl_python.rolling_batch.rolling_batch import filter_unused_generation_params
 from djl_python.request import Request
 from djl_python.transformers_neuronx_scheduler.token_selector import TokenSelector
 from djl_python.transformers_neuronx_scheduler.speculation import (

engines/python/setup/djl_python/transformers_neuronx_scheduler/optimum_token_selector.py

@@ -0,0 +1,222 @@
+#!/usr/bin/env python
+#
+# Copyright 2024 Amazon.com, Inc. or its affiliates. All Rights Reserved.
+#
+# Licensed under the Apache License, Version 2.0 (the "License"). You may not use this file
+# except in compliance with the License. A copy of the License is located at
+#
+# http://aws.amazon.com/apache2.0/
+#
+# or in the "LICENSE.txt" file accompanying this file. This file is distributed on an "AS IS"
+# BASIS, WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, express or implied. See the License for
+# the specific language governing permissions and limitations under the License.
+# The below code is heavily inspired from Optimum Neuron under the following link:
+# https://github.com/huggingface/optimum-neuron/blob/main/optimum/neuron/generation/token_selector.py
+
+import copy
+import logging
+from typing import TYPE_CHECKING, List, Optional
+
+import torch
+from transformers.generation import (
+    GenerationConfig,
+    GenerationMixin,
+    LogitsProcessorList,
+    StoppingCriteriaList,
+)
+from transformers.generation.utils import GenerationMode
+
+from optimum.neuron.generation import FusedLogitsWarper
+
+if TYPE_CHECKING:
+    from transformers import PreTrainedTokenizer
+
+logger = logging.getLogger(__name__)
+
+
+# TODO: This is a temporary solution to avoid Optimum's dependency on transformers<4.42.
+class OptimumTokenSelector:
+    """Implements the token selection logic corresponding to a generation configuration.
+
+    This class combines and uses the logits processors and stopping criterias implemented in
+    the transformers library.
+
+    The algorithm to select these objects is heavily inspired by the transformers `GenerationMixin.generate()`
+    method, but the actual token selection methods are specific.
+
+    The reason why this class does not inherit from `GenerationMixin` is because it does not
+    include the code to produce the tokens logits.
+    Separating the production of the tokens logits from the tokens selection allows this class
+    to be used with different generation paradigms, either synchronously using a single `TokenSelector` in
+    `GenerationMixin.generate()` or asynchronously using multiple `TokenSelector` inside an inference endpoint.
+
+    The constructor of this class should not be called directly: instances should be obtained by
+    calling `TokenSelector.create()`.
+    """
+
+    def __init__(
+        self,
+        mode: GenerationMode,
+        logits_processor: LogitsProcessorList,
+        stopping_criteria: StoppingCriteriaList,
+        eos_token_ids: List[int],
+        pad_token_id: int,
+        logits_warper: Optional[LogitsProcessorList] = None,
+        seed: Optional[int] = 0,
+    ):
+        self.mode = mode
+        self.logits_processor = logits_processor
+        self.stopping_criteria = stopping_criteria
+        self.eos_token_ids = eos_token_ids
+        self.pad_token_id = pad_token_id
+        self.logits_warper = logits_warper
+        self.generator = torch.Generator()
+        self.generator.manual_seed(seed)
+
+    @classmethod
+    def create(
+        cls,
+        input_ids: torch.Tensor,
+        generation_config: GenerationConfig,
+        model: GenerationMixin,
+        max_seq_length: int,
+        stopping_criteria: Optional[StoppingCriteriaList] = None,
+        tokenizer: Optional["PreTrainedTokenizer"] = None,
+        seed: Optional[int] = 0,
+    ) -> "OptimumTokenSelector":
+        r"""Creates the `TokenSelector` for a specific generation configuration.
+
+        Args:
+            input_ids (`torch.LongTensor` of shape `(batch_size, sequence_length)`):
+                The sequence used as a prompt for the generation.
+            generation_config (`~transformers.generation.GenerationConfig`, *optional*):
+                The generation configuration to parametrize the token selection.
+            model (`~transformers.generation.GenerationMixin`):
+                The model provides the internal helpers allowing to select the logits processors and stopping criterias.
+            max_seq_length (`int`):
+                The maximum number of input + generated tokens for this model. It depends on the model compilation parameters.
+            stopping_criteria (`Optional[transformers.generation.StoppingCriteriaList], defaults to `None`):
+                Custom stopping criteria that complement the default stopping criteria built from arguments and a
+                generation config
+            tokenizer (`Optional[transformers.PreTrainedTokenizer]`, default to `None`):
+                A tokenizer used when stop strings are passed to generate.
+            seed(`Optional[int]`):
+                The optional seed for sampling. Defaults to zero.
+        Return:
+            `torch.LongTensor`: A `torch.LongTensor` containing the selected tokens.
+        """
+        generation_config.validate()
+        generation_config = copy.deepcopy(generation_config)
+
+        unsupported_generation_flags = [
+            "output_attentions",
+            "output_hidden_states",
+            "output_scores",
+            "return_dict_in_generate",
+        ]
+        for flag in unsupported_generation_flags:
+            if getattr(generation_config, flag, False):
+                raise ValueError("{flag} is not supported for generation.")
+
+        if generation_config.max_new_tokens is not None:
+            logger.warning(
+                f"Both `max_new_tokens` (={generation_config.max_new_tokens}) and `max_length`(="
+                f"{generation_config.max_length}) seem to have been set. `max_new_tokens` will take precedence. "
+                "Please refer to the documentation for more information. "
+                "(https://huggingface.co/docs/transformers/main/en/main_classes/text_generation)"
+            )
+            generation_config.max_length = generation_config.max_new_tokens + input_ids.shape[
+                -1]
+
+        min_length = generation_config.min_length
+        if min_length > max_seq_length:
+            raise ValueError(
+                f"The minimum generation length ({min_length}) exceeds the model maximum sequence length ({max_seq_length})"
+            )
+        max_length = generation_config.max_length
+        if max_length > max_seq_length:
+            logger.warning(
+                f"Adjusting the maximum generation length ({max_length}) to the model maximum sequence length ({max_seq_length})"
+            )
+            generation_config.max_length = max_seq_length
+
+        # This is not supposed to happen for any of the models we support
+        eos_token_id = generation_config.eos_token_id
+        assert eos_token_id is not None
+        # The generation requires special tokens
+        eos_token_ids = eos_token_id if isinstance(eos_token_id,
+                                                   list) else [eos_token_id]
+        generation_config._eos_token_tensor = torch.tensor(
+            eos_token_ids, device=input_ids.device)
+        if generation_config.pad_token_id is None:
+            logger.warning(
+                f"Setting `pad_token_id` to `eos_token_id`:{eos_token_ids[0]} for open-ended generation."
+            )
+            generation_config.pad_token_id = eos_token_ids[0]
+
+        # Instantiate transformers library processors and criterias
+        logits_processor = model._get_logits_processor(
+            generation_config,
+            input_ids_seq_length=input_ids.shape[-1],
+            encoder_input_ids=input_ids,
+            prefix_allowed_tokens_fn=None,
+            logits_processor=LogitsProcessorList(),
+        )
+        if stopping_criteria is None:
+            stopping_criteria = StoppingCriteriaList()
+        stopping_criteria = model._get_stopping_criteria(
+            generation_config,
+            stopping_criteria=stopping_criteria,
+            tokenizer=tokenizer)
+
+        generation_mode = generation_config.get_generation_mode()
+        if generation_mode not in [
+                GenerationMode.GREEDY_SEARCH, GenerationMode.SAMPLE
+        ]:
+            raise ValueError("Unsupported generation mode")
+
+        logits_warper = None
+        if generation_mode == GenerationMode.SAMPLE:
+            logits_warper = FusedLogitsWarper.from_config(generation_config)
+
+        return cls(
+            mode=generation_mode,
+            logits_processor=logits_processor,
+            stopping_criteria=stopping_criteria,
+            logits_warper=logits_warper,
+            eos_token_ids=eos_token_ids,
+            pad_token_id=generation_config.pad_token_id,
+            seed=seed,
+        )
+
+    def select(self, input_ids: torch.LongTensor,
+               logits: torch.Tensor) -> torch.LongTensor:
+        """Select the next tokens from the candidate logits.
+
+        Args:
+            input_ids (`torch.LongTensor` of shape `(batch_size, sequence_length)`):
+                The sequence used as a prompt for the generation (not used in all generation modes).
+            logits (`torch.Tensor` of shape `(batch_size, sequence_length)`):
+                The logits corresponding to the generated tokens.
+
+        Return:
+            `torch.LongTensor`: A `torch.LongTensor` containing the selected tokens.
+        """
+        scores = self.logits_processor(input_ids, logits)
+        if self.mode == GenerationMode.SAMPLE:
+            return self._sample(scores)
+        else:
+            return torch.argmax(scores, dim=-1)
+
+    def _sample(self, scores: torch.Tensor) -> torch.LongTensor:
+        # Get [batch_size, kept] scores and indices instead of [batch_size, vocab_size] scores
+        scores, next_token_indices = self.logits_warper(scores)
+
+        # sample
+        probs = torch.nn.functional.softmax(scores, dim=-1)
+        next_tokens = torch.multinomial(probs,
+                                        num_samples=1,
+                                        generator=self.generator)
+        # Convert the filtered tokens to actual vocabulary tokens
+        next_tokens = torch.gather(next_token_indices, 1, next_tokens)
+        return next_tokens.squeeze(1)

engines/python/setup/djl_python/transformers_neuronx_scheduler/token_selector.py

@@ -169,7 +169,8 @@ def create(
 
         logits_warper = None
         if generation_mode == GenerationMode.SAMPLE:
-            logits_warper = model._get_logits_warper(generation_config)
+            logits_warper = model._get_logits_warper(generation_config,
+                                                     device=model.device)
             if len(logits_warper) == 0:
                 generation_mode = GenerationMode.GREEDY_SEARCH