From e4967a564e3f215cd6f76f67b6722d8c5ee25fb5 Mon Sep 17 00:00:00 2001
From: Callum McLean <callumm@graphcore.ai>
Date: Tue, 20 Jun 2023 16:40:52 +0100
Subject: [PATCH 1/8] Add `run_speech_recognition_seq2seq.py`

---
 .../run_speech_recognition_seq2seq.py         | 618 ++++++++++++++++++
 1 file changed, 618 insertions(+)
 create mode 100755 examples/speech-recognition/run_speech_recognition_seq2seq.py

diff --git a/examples/speech-recognition/run_speech_recognition_seq2seq.py b/examples/speech-recognition/run_speech_recognition_seq2seq.py
new file mode 100755
index 000000000..12d82b86e
--- /dev/null
+++ b/examples/speech-recognition/run_speech_recognition_seq2seq.py
@@ -0,0 +1,618 @@
+#!/usr/bin/env python
+# coding=utf-8
+# Copyright 2021 The HuggingFace Team. 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.
+# You may obtain a copy of the License at
+#
+#     http://www.apache.org/licenses/LICENSE-2.0
+#
+# Unless required by applicable law or agreed to in writing, software
+# distributed under the License is distributed on an "AS IS" BASIS,
+# WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+# See the License for the specific language governing permissions and
+# limitations under the License.
+"""
+Fine-tuning the library models for sequence to sequence speech recognition.
+"""
+# You can also adapt this script on your own sequence to sequence speech
+# recognition task. Pointers for this are left as comments.
+
+import logging
+import os
+import sys
+import warnings
+from dataclasses import dataclass, field
+from typing import Any, Dict, List, Optional, Union
+
+import datasets
+import evaluate
+import torch
+from datasets import DatasetDict, load_dataset
+
+import transformers
+from transformers import (
+    AutoConfig,
+    AutoFeatureExtractor,
+    AutoModelForSpeechSeq2Seq,
+    AutoProcessor,
+    AutoTokenizer,
+    HfArgumentParser,
+    WhisperProcessor,
+    set_seed,
+)
+from transformers.trainer_utils import get_last_checkpoint
+from transformers.utils import check_min_version, send_example_telemetry
+from transformers.utils.versions import require_version
+
+from optimum.graphcore import IPUConfig, IPUSeq2SeqTrainer
+from optimum.graphcore import IPUSeq2SeqTrainingArguments as Seq2SeqTrainingArguments
+
+# Will error if the minimal version of Transformers is not installed. Remove at your own risks.
+check_min_version("4.29.0")
+
+require_version("datasets>=1.18.0", "To fix: pip install -r examples/pytorch/speech-recognition/requirements.txt")
+
+logger = logging.getLogger(__name__)
+
+
+@dataclass
+class ModelArguments:
+    """
+    Arguments pertaining to which model/config/tokenizer we are going to fine-tune from.
+    """
+
+    model_name_or_path: str = field(
+        metadata={"help": "Path to pretrained model or model identifier from huggingface.co/models"}
+    )
+    config_name: Optional[str] = field(
+        default=None, metadata={"help": "Pretrained config name or path if not the same as model_name"}
+    )
+    tokenizer_name: Optional[str] = field(
+        default=None, metadata={"help": "Pretrained tokenizer name or path if not the same as model_name"}
+    )
+    feature_extractor_name: Optional[str] = field(
+        default=None, metadata={"help": "feature extractor name or path if not the same as model_name"}
+    )
+    cache_dir: Optional[str] = field(
+        default=None,
+        metadata={"help": "Where to store the pretrained models downloaded from huggingface.co"},
+    )
+    use_fast_tokenizer: bool = field(
+        default=True,
+        metadata={"help": "Whether to use one of the fast tokenizer (backed by the tokenizers library) or not."},
+    )
+    model_revision: str = field(
+        default="main",
+        metadata={"help": "The specific model version to use (can be a branch name, tag name or commit id)."},
+    )
+    use_auth_token: bool = field(
+        default=False,
+        metadata={
+            "help": (
+                "Will use the token generated when running `huggingface-cli login` (necessary to use this script "
+                "with private models)."
+            )
+        },
+    )
+    freeze_feature_encoder: bool = field(
+        default=True, metadata={"help": "Whether to freeze the feature encoder layers of the model."}
+    )
+    freeze_encoder: bool = field(
+        default=False, metadata={"help": "Whether to freeze the entire encoder of the seq2seq model."}
+    )
+    forced_decoder_ids: List[List[int]] = field(
+        default=None,
+        metadata={
+            "help": (
+                "A list of pairs of integers which indicates a mapping from generation indices to token indices "
+                "that will be forced before sampling. For example, [[0, 123]] means the first generated token "
+                "will always be a token of index 123."
+            )
+        },
+    )
+    suppress_tokens: List[int] = field(
+        default=None, metadata={"help": "A list of tokens that will be suppressed at generation."}
+    )
+    apply_spec_augment: bool = field(
+        default=False,
+        metadata={
+            "help": "Whether to apply *SpecAugment* data augmentation to the input features. This is currently only relevant for Wav2Vec2, HuBERT, WavLM and Whisper models."
+        },
+    )
+
+
+@dataclass
+class DataTrainingArguments:
+    """
+    Arguments pertaining to what data we are going to input our model for training and eval.
+    """
+
+    dataset_name: str = field(
+        default=None, metadata={"help": "The name of the dataset to use (via the datasets library)."}
+    )
+    dataset_config_name: Optional[str] = field(
+        default=None, metadata={"help": "The configuration name of the dataset to use (via the datasets library)."}
+    )
+    overwrite_cache: bool = field(
+        default=False, metadata={"help": "Overwrite the cached training and evaluation sets"}
+    )
+    preprocessing_num_workers: Optional[int] = field(
+        default=None,
+        metadata={"help": "The number of processes to use for the preprocessing."},
+    )
+    max_train_samples: Optional[int] = field(
+        default=None,
+        metadata={
+            "help": (
+                "For debugging purposes or quicker training, truncate the number of training examples to this "
+                "value if set."
+            )
+        },
+    )
+    max_eval_samples: Optional[int] = field(
+        default=None,
+        metadata={
+            "help": (
+                "For debugging purposes or quicker training, truncate the number of evaluation examples to this "
+                "value if set."
+            )
+        },
+    )
+    audio_column_name: str = field(
+        default="audio",
+        metadata={"help": "The name of the dataset column containing the audio data. Defaults to 'audio'"},
+    )
+    text_column_name: str = field(
+        default="text",
+        metadata={"help": "The name of the dataset column containing the text data. Defaults to 'text'"},
+    )
+    max_duration_in_seconds: float = field(
+        default=20.0,
+        metadata={
+            "help": (
+                "Truncate audio files that are longer than `max_duration_in_seconds` seconds to"
+                " 'max_duration_in_seconds`"
+            )
+        },
+    )
+    min_duration_in_seconds: float = field(
+        default=0.0, metadata={"help": "Filter audio files that are shorter than `min_duration_in_seconds` seconds"}
+    )
+    preprocessing_only: bool = field(
+        default=False,
+        metadata={
+            "help": (
+                "Whether to only do data preprocessing and skip training. This is especially useful when data"
+                " preprocessing errors out in distributed training due to timeout. In this case, one should run the"
+                " preprocessing in a non-distributed setup with `preprocessing_only=True` so that the cached datasets"
+                " can consequently be loaded in distributed training"
+            )
+        },
+    )
+    train_split_name: str = field(
+        default="train",
+        metadata={
+            "help": "The name of the training data set split to use (via the datasets library). Defaults to 'train'"
+        },
+    )
+    eval_split_name: str = field(
+        default="test",
+        metadata={
+            "help": "The name of the training data set split to use (via the datasets library). Defaults to 'train'"
+        },
+    )
+    do_lower_case: bool = field(
+        default=True,
+        metadata={"help": "Whether the target text should be lower cased."},
+    )
+    language: str = field(
+        default=None,
+        metadata={
+            "help": (
+                "Language for multilingual fine-tuning. This argument should be set for multilingual fine-tuning "
+                "only. For English speech recognition, it should be set to `None`."
+            )
+        },
+    )
+    task: str = field(
+        default="transcribe",
+        metadata={"help": "Task, either `transcribe` for speech recognition or `translate` for speech translation."},
+    )
+
+
+@dataclass
+class DataCollatorSpeechSeq2SeqWithPadding:
+    """
+    Data collator that will dynamically pad the inputs received.
+    Args:
+        processor ([`WhisperProcessor`])
+            The processor used for processing the data.
+        decoder_start_token_id (`int`)
+            The begin-of-sentence of the decoder.
+        forward_attention_mask (`bool`)
+            Whether to return attention_mask.
+    """
+
+    processor: Any
+    decoder_start_token_id: int
+    forward_attention_mask: bool
+    padding: Union[bool, str] = "longest"
+    pad_to_multiple_of: Optional[int] = None
+    pad_to_multiple_of_labels: Optional[int] = None
+
+    def __call__(self, features: List[Dict[str, Union[List[int], torch.Tensor]]]) -> Dict[str, torch.Tensor]:
+        # split inputs and labels since they have to be of different lengths and need
+        # different padding methods
+        model_input_name = self.processor.model_input_names[0]
+        input_features = [{model_input_name: feature[model_input_name]} for feature in features]
+        label_features = [{"input_ids": feature["labels"]} for feature in features]
+
+        batch = self.processor.feature_extractor.pad(
+            input_features,
+            pad_to_multiple_of=self.pad_to_multiple_of,
+            return_tensors="pt"
+        )
+
+        if self.forward_attention_mask:
+            batch["attention_mask"] = torch.LongTensor([feature["attention_mask"] for feature in features])
+
+        labels_batch = self.processor.tokenizer.pad(
+            label_features,
+            pad_to_multiple_of=self.pad_to_multiple_of_labels,
+            return_tensors="pt"
+        )
+
+        # replace padding with -100 to ignore loss correctly
+        labels = labels_batch["input_ids"].masked_fill(labels_batch.attention_mask.ne(1), -100)
+
+        # if bos token is appended in previous tokenization step,
+        # cut bos token here as it's append later anyways
+        if (labels[:, 0] == self.decoder_start_token_id).all().cpu().item():
+            labels = labels[:, 1:]
+
+        batch["labels"] = labels
+
+        return batch.data
+
+
+def main():
+    # 1. Parse input arguments
+    # See all possible arguments in src/transformers/training_args.py
+    # or by passing the --help flag to this script.
+    # We now keep distinct sets of args, for a cleaner separation of concerns.
+    parser = HfArgumentParser((ModelArguments, DataTrainingArguments, Seq2SeqTrainingArguments))
+
+    if len(sys.argv) == 2 and sys.argv[1].endswith(".json"):
+        # If we pass only one argument to the script and it's the path to a json file,
+        # let's parse it to get our arguments.
+        model_args, data_args, training_args = parser.parse_json_file(json_file=os.path.abspath(sys.argv[1]))
+    else:
+        model_args, data_args, training_args = parser.parse_args_into_dataclasses()
+
+    if training_args.gradient_checkpointing:
+        raise ValueError("Gradient checkpointing not supported.")
+
+    # Sending telemetry. Tracking the example usage helps us better allocate resources to maintain them. The
+    # information sent is the one passed as arguments along with your Python/PyTorch versions.
+    send_example_telemetry("run_speech_recognition_seq2seq", model_args, data_args)
+
+    # 2. Setup logging
+    logging.basicConfig(
+        format="%(asctime)s - %(levelname)s - %(name)s - %(message)s",
+        datefmt="%m/%d/%Y %H:%M:%S",
+        handlers=[logging.StreamHandler(sys.stdout)],
+    )
+    log_level = training_args.get_process_log_level()
+    logger.setLevel(log_level)
+    datasets.utils.logging.set_verbosity(log_level)
+    transformers.utils.logging.set_verbosity(log_level)
+    transformers.utils.logging.enable_default_handler()
+    transformers.utils.logging.enable_explicit_format()
+
+    # Log on each process the small summary:
+    logger.warning(
+        f"Process rank: {training_args.local_rank}, device: {training_args.device}, n_gpu: {training_args.n_gpu}"
+        f"distributed training: {bool(training_args.local_rank != -1)}, 16-bits training: {training_args.fp16}"
+    )
+    logger.info(f"Training/evaluation parameters {training_args}")
+
+    # 3. Detecting last checkpoint and eventually continue from last checkpoint
+    last_checkpoint = None
+    if os.path.isdir(training_args.output_dir) and training_args.do_train and not training_args.overwrite_output_dir:
+        last_checkpoint = get_last_checkpoint(training_args.output_dir)
+        if last_checkpoint is None and len(os.listdir(training_args.output_dir)) > 0:
+            raise ValueError(
+                f"Output directory ({training_args.output_dir}) already exists and is not empty. "
+                "Use --overwrite_output_dir to overcome."
+            )
+        elif last_checkpoint is not None and training_args.resume_from_checkpoint is None:
+            logger.info(
+                f"Checkpoint detected, resuming training at {last_checkpoint}. To avoid this behavior, change "
+                "the `--output_dir` or add `--overwrite_output_dir` to train from scratch."
+            )
+
+    # Set seed before initializing model.
+    set_seed(training_args.seed)
+
+    # 4. Load dataset
+    raw_datasets = DatasetDict()
+
+    if training_args.do_train:
+        raw_datasets["train"] = load_dataset(
+            data_args.dataset_name,
+            data_args.dataset_config_name,
+            split=data_args.train_split_name,
+            cache_dir=model_args.cache_dir,
+            use_auth_token=True if model_args.use_auth_token else None,
+        )
+
+    if training_args.do_eval:
+        raw_datasets["eval"] = load_dataset(
+            data_args.dataset_name,
+            data_args.dataset_config_name,
+            split=data_args.eval_split_name,
+            cache_dir=model_args.cache_dir,
+            use_auth_token=True if model_args.use_auth_token else None,
+        )
+
+    if data_args.audio_column_name not in next(iter(raw_datasets.values())).column_names:
+        raise ValueError(
+            f"--audio_column_name '{data_args.audio_column_name}' not found in dataset '{data_args.dataset_name}'. "
+            "Make sure to set `--audio_column_name` to the correct audio column - one of "
+            f"{', '.join(next(iter(raw_datasets.values())).column_names)}."
+        )
+
+    if data_args.text_column_name not in next(iter(raw_datasets.values())).column_names:
+        raise ValueError(
+            f"--text_column_name {data_args.text_column_name} not found in dataset '{data_args.dataset_name}'. "
+            "Make sure to set `--text_column_name` to the correct text column - one of "
+            f"{', '.join(next(iter(raw_datasets.values())).column_names)}."
+        )
+
+    # 5. Load pretrained model, tokenizer, and feature extractor
+    #
+    # Distributed training:
+    # The .from_pretrained methods guarantee that only one local process can concurrently
+    config = AutoConfig.from_pretrained(
+        model_args.config_name if model_args.config_name else model_args.model_name_or_path,
+        cache_dir=model_args.cache_dir,
+        revision=model_args.model_revision,
+        use_auth_token=True if model_args.use_auth_token else None,
+    )
+
+    config.update({"forced_decoder_ids": model_args.forced_decoder_ids, "suppress_tokens": model_args.suppress_tokens})
+
+    # SpecAugment for whisper models
+    if getattr(config, "model_type", None) == "whisper":
+        if model_args.apply_spec_augment:
+            raise ValueError("SpecAugment is not supported on IPU")
+        config.update({"apply_spec_augment": model_args.apply_spec_augment})
+
+    # IPU specific config updates
+    config.update({"apply_spec_augment": False})
+        
+    # Whisper does not have a layer_norm_eps option, remains to be seen if this is a problem
+    #config.update({"layer_norm_eps": 0.0001})
+
+    feature_extractor = AutoFeatureExtractor.from_pretrained(
+        model_args.feature_extractor_name if model_args.feature_extractor_name else model_args.model_name_or_path,
+        cache_dir=model_args.cache_dir,
+        revision=model_args.model_revision,
+        use_auth_token=True if model_args.use_auth_token else None,
+    )
+    tokenizer = AutoTokenizer.from_pretrained(
+        model_args.tokenizer_name if model_args.tokenizer_name else model_args.model_name_or_path,
+        cache_dir=model_args.cache_dir,
+        use_fast=model_args.use_fast_tokenizer,
+        revision=model_args.model_revision,
+        use_auth_token=True if model_args.use_auth_token else None,
+    )
+    model = AutoModelForSpeechSeq2Seq.from_pretrained(
+        model_args.model_name_or_path,
+        config=config,
+        cache_dir=model_args.cache_dir,
+        revision=model_args.model_revision,
+        use_auth_token=True if model_args.use_auth_token else None,
+    )
+    ipu_config = IPUConfig.from_pretrained(
+        training_args.ipu_config_name if training_args.ipu_config_name else model_args.model_name_or_path,
+        cache_dir=model_args.cache_dir,
+        use_auth_token=True if data_args.use_auth_token else None,
+    )
+    if model.config.decoder_start_token_id is None:
+        raise ValueError("Make sure that `config.decoder_start_token_id` is correctly defined")
+
+    if model_args.freeze_feature_encoder:
+        model.freeze_feature_encoder()
+
+    if model_args.freeze_encoder:
+        model.freeze_encoder()
+        model.model.encoder.gradient_checkpointing = False
+
+    if data_args.language is not None:
+        # We only need to set the task id when the language is specified (i.e. in a multilingual setting)
+        tokenizer.set_prefix_tokens(language=data_args.language, task=data_args.task)
+
+    # 6. Resample speech dataset if necessary
+    dataset_sampling_rate = next(iter(raw_datasets.values())).features[data_args.audio_column_name].sampling_rate
+    if dataset_sampling_rate != feature_extractor.sampling_rate:
+        raw_datasets = raw_datasets.cast_column(
+            data_args.audio_column_name, datasets.features.Audio(sampling_rate=feature_extractor.sampling_rate)
+        )
+
+    # 7. Preprocessing the datasets.
+    # We need to read the audio files as arrays and tokenize the targets.
+    max_input_length = data_args.max_duration_in_seconds * feature_extractor.sampling_rate
+    min_input_length = data_args.min_duration_in_seconds * feature_extractor.sampling_rate
+    audio_column_name = data_args.audio_column_name
+    num_workers = data_args.preprocessing_num_workers
+    text_column_name = data_args.text_column_name
+    model_input_name = feature_extractor.model_input_names[0]
+    do_lower_case = data_args.do_lower_case
+    # if SpecAugment is used for whisper models, return attention_mask to guide the mask along time axis
+    forward_attention_mask = (
+        getattr(config, "model_type", None) == "whisper"
+        and getattr(config, "apply_spec_augment", False)
+        and getattr(config, "mask_time_prob", 0) > 0
+    )
+
+    if data_args.max_train_samples is not None:
+        raw_datasets["train"] = raw_datasets["train"].select(range(data_args.max_train_samples))
+
+    if data_args.max_eval_samples is not None:
+        raw_datasets["eval"] = raw_datasets["eval"].select(range(data_args.max_eval_samples))
+
+    def prepare_dataset(batch, feature_extractor, tokenizer):
+        # process audio
+        sample = batch[audio_column_name]
+        inputs = feature_extractor(
+            sample["array"], sampling_rate=sample["sampling_rate"], return_attention_mask=forward_attention_mask
+        )
+        # process audio length
+        batch[model_input_name] = inputs.get(model_input_name)[0]
+        batch["input_length"] = len(sample["array"])
+        if forward_attention_mask:
+            batch["attention_mask"] = inputs.get("attention_mask")[0]
+
+        if not training_args.fp32:
+            # Cast audio inputs to FP16
+            batch["input_values"] = batch["input_values"].astype(np.float16)            
+            
+        # process targets
+        input_str = batch[text_column_name].lower() if do_lower_case else batch[text_column_name]
+        batch["labels"] = tokenizer(input_str).input_ids
+        return batch
+
+    with training_args.main_process_first(desc="dataset map pre-processing"):
+        vectorized_datasets = raw_datasets.map(
+            lambda batch: prepare_dataset(batch, feature_extractor, tokenizer),
+            remove_columns=next(iter(raw_datasets.values())).column_names,
+            num_proc=data_args.preprocessing_num_workers,
+            desc="preprocess train dataset",
+        )
+
+    # filter data that is shorter than min_input_length or longer than
+    # max_input_length
+    def is_audio_in_length_range(length):
+        return length > min_input_length and length < max_input_length
+
+    vectorized_datasets = vectorized_datasets.filter(
+        is_audio_in_length_range,
+        num_proc=num_workers,
+        input_columns=["input_length"],
+    )
+
+    # for large datasets it is advised to run the preprocessing on a
+    # single machine first with `args.preprocessing_only` since there will mostly likely
+    # be a timeout when running the script in distributed mode.
+    # In a second step `args.preprocessing_only` can then be set to `False` to load the
+    # cached dataset
+    if data_args.preprocessing_only:
+        cache = {k: v.cache_files for k, v in vectorized_datasets.items()}
+        logger.info(f"Data preprocessing finished. Files cached at {cache}.")
+        return
+
+    # 8. Load Metric
+    metric = evaluate.load("wer")
+
+    def compute_metrics(pred):
+        pred_ids = pred.predictions
+
+        pred.label_ids[pred.label_ids == -100] = tokenizer.pad_token_id
+
+        pred_str = tokenizer.batch_decode(pred_ids, skip_special_tokens=True)
+        # we do not want to group tokens when computing the metrics
+        label_str = tokenizer.batch_decode(pred.label_ids, skip_special_tokens=True)
+
+        wer = metric.compute(predictions=pred_str, references=label_str)
+
+        return {"wer": wer}
+
+    # 9. Create a single speech processor
+    # save feature extractor, tokenizer and config
+    feature_extractor.save_pretrained(training_args.output_dir)
+    tokenizer.save_pretrained(training_args.output_dir)
+    config.save_pretrained(training_args.output_dir)
+
+    processor = WhisperProcessor(feature_extractor, tokenizer)
+
+    # 10. Define data collator
+    data_collator = DataCollatorSpeechSeq2SeqWithPadding(
+        processor=processor,
+        decoder_start_token_id=model.config.decoder_start_token_id,
+        forward_attention_mask=forward_attention_mask,
+        pad_to_multiple_of=max_input_length,
+        pad_to_multiple_of_labels=500
+    )
+
+    # 11. Initialize Trainer
+    trainer = IPUSeq2SeqTrainer(
+        model=model,
+        ipu_config=ipu_config,
+        args=training_args,
+        train_dataset=vectorized_datasets["train"] if training_args.do_train else None,
+        eval_dataset=vectorized_datasets["eval"] if training_args.do_eval else None,
+        data_collator=data_collator,
+        compute_metrics=compute_metrics if training_args.predict_with_generate else None,
+    )
+
+    # 12. Training
+    if training_args.do_train:
+        checkpoint = None
+        if training_args.resume_from_checkpoint is not None:
+            checkpoint = training_args.resume_from_checkpoint
+        elif last_checkpoint is not None:
+            checkpoint = last_checkpoint
+        train_result = trainer.train(resume_from_checkpoint=checkpoint)
+        trainer.save_model()  # Saves the feature extractor too for easy upload
+
+        metrics = train_result.metrics
+        max_train_samples = (
+            data_args.max_train_samples
+            if data_args.max_train_samples is not None
+            else len(vectorized_datasets["train"])
+        )
+        metrics["train_samples"] = min(max_train_samples, len(vectorized_datasets["train"]))
+        trainer.log_metrics("train", metrics)
+        trainer.save_metrics("train", metrics)
+        trainer.save_state()
+
+    # 13. Evaluation
+    results = {}
+    if training_args.do_eval:
+        logger.info("*** Evaluate ***")
+        metrics = trainer.evaluate(
+            metric_key_prefix="eval",
+            max_length=training_args.generation_max_length,
+            num_beams=training_args.generation_num_beams,
+        )
+        max_eval_samples = (
+            data_args.max_eval_samples if data_args.max_eval_samples is not None else len(vectorized_datasets["eval"])
+        )
+        metrics["eval_samples"] = min(max_eval_samples, len(vectorized_datasets["eval"]))
+
+        trainer.log_metrics("eval", metrics)
+        trainer.save_metrics("eval", metrics)
+
+    # 14. Write Training Stats
+    kwargs = {"finetuned_from": model_args.model_name_or_path, "tasks": "automatic-speech-recognition"}
+    if data_args.dataset_name is not None:
+        kwargs["dataset_tags"] = data_args.dataset_name
+        if data_args.dataset_config_name is not None:
+            kwargs["dataset_args"] = data_args.dataset_config_name
+            kwargs["dataset"] = f"{data_args.dataset_name} {data_args.dataset_config_name}"
+        else:
+            kwargs["dataset"] = data_args.dataset_name
+
+    if training_args.push_to_hub:
+        trainer.push_to_hub(**kwargs)
+    else:
+        trainer.create_model_card(**kwargs)
+
+    return results
+
+
+if __name__ == "__main__":
+    main()

From 6e01df2977d717eef3f5681539cb479a3e531a3c Mon Sep 17 00:00:00 2001
From: Callum McLean <callumm@graphcore.ai>
Date: Tue, 4 Jul 2023 16:32:21 +0100
Subject: [PATCH 2/8] Script now working with interleaved training + validation

---
 .../run_speech_recognition_seq2seq.py         | 36 ++++++++++---------
 1 file changed, 20 insertions(+), 16 deletions(-)

diff --git a/examples/speech-recognition/run_speech_recognition_seq2seq.py b/examples/speech-recognition/run_speech_recognition_seq2seq.py
index 12d82b86e..692aa0286 100755
--- a/examples/speech-recognition/run_speech_recognition_seq2seq.py
+++ b/examples/speech-recognition/run_speech_recognition_seq2seq.py
@@ -20,6 +20,7 @@
 # recognition task. Pointers for this are left as comments.
 
 import logging
+import math
 import os
 import sys
 import warnings
@@ -28,6 +29,7 @@
 
 import datasets
 import evaluate
+import numpy as np
 import torch
 from datasets import DatasetDict, load_dataset
 
@@ -312,10 +314,6 @@ def main():
     transformers.utils.logging.enable_explicit_format()
 
     # Log on each process the small summary:
-    logger.warning(
-        f"Process rank: {training_args.local_rank}, device: {training_args.device}, n_gpu: {training_args.n_gpu}"
-        f"distributed training: {bool(training_args.local_rank != -1)}, 16-bits training: {training_args.fp16}"
-    )
     logger.info(f"Training/evaluation parameters {training_args}")
 
     # 3. Detecting last checkpoint and eventually continue from last checkpoint
@@ -419,8 +417,9 @@ def main():
     ipu_config = IPUConfig.from_pretrained(
         training_args.ipu_config_name if training_args.ipu_config_name else model_args.model_name_or_path,
         cache_dir=model_args.cache_dir,
-        use_auth_token=True if data_args.use_auth_token else None,
+        use_auth_token=True if model_args.use_auth_token else None,
     )
+
     if model.config.decoder_start_token_id is None:
         raise ValueError("Make sure that `config.decoder_start_token_id` is correctly defined")
 
@@ -470,7 +469,7 @@ def prepare_dataset(batch, feature_extractor, tokenizer):
         inputs = feature_extractor(
             sample["array"], sampling_rate=sample["sampling_rate"], return_attention_mask=forward_attention_mask
         )
-        # process audio length
+
         batch[model_input_name] = inputs.get(model_input_name)[0]
         batch["input_length"] = len(sample["array"])
         if forward_attention_mask:
@@ -478,20 +477,19 @@ def prepare_dataset(batch, feature_extractor, tokenizer):
 
         if not training_args.fp32:
             # Cast audio inputs to FP16
-            batch["input_values"] = batch["input_values"].astype(np.float16)            
+            batch[model_input_name] = batch[model_input_name].astype(np.float16)            
             
         # process targets
         input_str = batch[text_column_name].lower() if do_lower_case else batch[text_column_name]
         batch["labels"] = tokenizer(input_str).input_ids
         return batch
 
-    with training_args.main_process_first(desc="dataset map pre-processing"):
-        vectorized_datasets = raw_datasets.map(
-            lambda batch: prepare_dataset(batch, feature_extractor, tokenizer),
-            remove_columns=next(iter(raw_datasets.values())).column_names,
-            num_proc=data_args.preprocessing_num_workers,
-            desc="preprocess train dataset",
-        )
+    vectorized_datasets = raw_datasets.map(
+        lambda batch: prepare_dataset(batch, feature_extractor, tokenizer),
+        remove_columns=next(iter(raw_datasets.values())).column_names,
+        num_proc=data_args.preprocessing_num_workers,
+        desc="preprocess train dataset",
+    )
 
     # filter data that is shorter than min_input_length or longer than
     # max_input_length
@@ -543,8 +541,9 @@ def compute_metrics(pred):
         processor=processor,
         decoder_start_token_id=model.config.decoder_start_token_id,
         forward_attention_mask=forward_attention_mask,
-        pad_to_multiple_of=max_input_length,
-        pad_to_multiple_of_labels=500
+        #pad_to_multiple_of=math.ceil(max_input_length),
+        pad_to_multiple_of=80,
+        pad_to_multiple_of_labels=training_args.generation_max_length
     )
 
     # 11. Initialize Trainer
@@ -556,6 +555,11 @@ def compute_metrics(pred):
         eval_dataset=vectorized_datasets["eval"] if training_args.do_eval else None,
         data_collator=data_collator,
         compute_metrics=compute_metrics if training_args.predict_with_generate else None,
+        eval_parallelize_kwargs={
+            'use_cache': True,
+            'use_cross_cache': True,
+            'max_length': training_args.generation_max_length
+        }
     )
 
     # 12. Training

From 5ac00cab369ebcc87ccd863cd42b9f79e2d0ed23 Mon Sep 17 00:00:00 2001
From: Callum McLean <callumm@graphcore.ai>
Date: Tue, 11 Jul 2023 17:33:26 +0100
Subject: [PATCH 3/8] `make style` + add diff .txt to tests

---
 .../run_speech_recognition_seq2seq.py         |  35 ++--
 .../run_speech_recognition_seq2seq.txt        | 151 ++++++++++++++++++
 2 files changed, 165 insertions(+), 21 deletions(-)
 create mode 100644 tests/examples/run_speech_recognition_seq2seq.txt

diff --git a/examples/speech-recognition/run_speech_recognition_seq2seq.py b/examples/speech-recognition/run_speech_recognition_seq2seq.py
index 692aa0286..9ba32c4ca 100755
--- a/examples/speech-recognition/run_speech_recognition_seq2seq.py
+++ b/examples/speech-recognition/run_speech_recognition_seq2seq.py
@@ -20,10 +20,8 @@
 # recognition task. Pointers for this are left as comments.
 
 import logging
-import math
 import os
 import sys
-import warnings
 from dataclasses import dataclass, field
 from typing import Any, Dict, List, Optional, Union
 
@@ -31,14 +29,12 @@
 import evaluate
 import numpy as np
 import torch
-from datasets import DatasetDict, load_dataset
-
 import transformers
+from datasets import DatasetDict, load_dataset
 from transformers import (
     AutoConfig,
     AutoFeatureExtractor,
     AutoModelForSpeechSeq2Seq,
-    AutoProcessor,
     AutoTokenizer,
     HfArgumentParser,
     WhisperProcessor,
@@ -51,6 +47,7 @@
 from optimum.graphcore import IPUConfig, IPUSeq2SeqTrainer
 from optimum.graphcore import IPUSeq2SeqTrainingArguments as Seq2SeqTrainingArguments
 
+
 # Will error if the minimal version of Transformers is not installed. Remove at your own risks.
 check_min_version("4.29.0")
 
@@ -252,18 +249,14 @@ def __call__(self, features: List[Dict[str, Union[List[int], torch.Tensor]]]) ->
         label_features = [{"input_ids": feature["labels"]} for feature in features]
 
         batch = self.processor.feature_extractor.pad(
-            input_features,
-            pad_to_multiple_of=self.pad_to_multiple_of,
-            return_tensors="pt"
+            input_features, pad_to_multiple_of=self.pad_to_multiple_of, return_tensors="pt"
         )
 
         if self.forward_attention_mask:
             batch["attention_mask"] = torch.LongTensor([feature["attention_mask"] for feature in features])
 
         labels_batch = self.processor.tokenizer.pad(
-            label_features,
-            pad_to_multiple_of=self.pad_to_multiple_of_labels,
-            return_tensors="pt"
+            label_features, pad_to_multiple_of=self.pad_to_multiple_of_labels, return_tensors="pt"
         )
 
         # replace padding with -100 to ignore loss correctly
@@ -390,9 +383,9 @@ def main():
 
     # IPU specific config updates
     config.update({"apply_spec_augment": False})
-        
+
     # Whisper does not have a layer_norm_eps option, remains to be seen if this is a problem
-    #config.update({"layer_norm_eps": 0.0001})
+    # config.update({"layer_norm_eps": 0.0001})
 
     feature_extractor = AutoFeatureExtractor.from_pretrained(
         model_args.feature_extractor_name if model_args.feature_extractor_name else model_args.model_name_or_path,
@@ -477,8 +470,8 @@ def prepare_dataset(batch, feature_extractor, tokenizer):
 
         if not training_args.fp32:
             # Cast audio inputs to FP16
-            batch[model_input_name] = batch[model_input_name].astype(np.float16)            
-            
+            batch[model_input_name] = batch[model_input_name].astype(np.float16)
+
         # process targets
         input_str = batch[text_column_name].lower() if do_lower_case else batch[text_column_name]
         batch["labels"] = tokenizer(input_str).input_ids
@@ -541,9 +534,9 @@ def compute_metrics(pred):
         processor=processor,
         decoder_start_token_id=model.config.decoder_start_token_id,
         forward_attention_mask=forward_attention_mask,
-        #pad_to_multiple_of=math.ceil(max_input_length),
+        # pad_to_multiple_of=math.ceil(max_input_length),
         pad_to_multiple_of=80,
-        pad_to_multiple_of_labels=training_args.generation_max_length
+        pad_to_multiple_of_labels=training_args.generation_max_length,
     )
 
     # 11. Initialize Trainer
@@ -556,10 +549,10 @@ def compute_metrics(pred):
         data_collator=data_collator,
         compute_metrics=compute_metrics if training_args.predict_with_generate else None,
         eval_parallelize_kwargs={
-            'use_cache': True,
-            'use_cross_cache': True,
-            'max_length': training_args.generation_max_length
-        }
+            "use_cache": True,
+            "use_cross_cache": True,
+            "max_length": training_args.generation_max_length,
+        },
     )
 
     # 12. Training
diff --git a/tests/examples/run_speech_recognition_seq2seq.txt b/tests/examples/run_speech_recognition_seq2seq.txt
new file mode 100644
index 000000000..81d3cb0b9
--- /dev/null
+++ b/tests/examples/run_speech_recognition_seq2seq.txt
@@ -0,0 +1,151 @@
+29a30
+> import numpy as np
+31,32d31
+< from datasets import DatasetDict, load_dataset
+< 
+33a33
+> from datasets import DatasetDict, load_dataset
+38d37
+<     AutoProcessor,
+41,42c40
+<     Seq2SeqTrainer,
+<     Seq2SeqTrainingArguments,
+---
+>     WhisperProcessor,
+45c43
+< from transformers.trainer_utils import get_last_checkpoint, is_main_process
+---
+> from transformers.trainer_utils import get_last_checkpoint
+48a47,49
+> from optimum.graphcore import IPUConfig, IPUSeq2SeqTrainer
+> from optimum.graphcore import IPUSeq2SeqTrainingArguments as Seq2SeqTrainingArguments
+> 
+51c52
+< check_min_version("4.31.0.dev0")
+---
+> check_min_version("4.29.0")
+238a240,242
+>     padding: Union[bool, str] = "longest"
+>     pad_to_multiple_of: Optional[int] = None
+>     pad_to_multiple_of_labels: Optional[int] = None
+247c251,253
+<         batch = self.processor.feature_extractor.pad(input_features, return_tensors="pt")
+---
+>         batch = self.processor.feature_extractor.pad(
+>             input_features, pad_to_multiple_of=self.pad_to_multiple_of, return_tensors="pt"
+>         )
+252c258,260
+<         labels_batch = self.processor.tokenizer.pad(label_features, return_tensors="pt")
+---
+>         labels_batch = self.processor.tokenizer.pad(
+>             label_features, pad_to_multiple_of=self.pad_to_multiple_of_labels, return_tensors="pt"
+>         )
+264c272
+<         return batch
+---
+>         return batch.data
+280a289,294
+>     if training_args.gradient_checkpointing:
+>         raise ValueError("Gradient checkpointing not supported.")
+> 
+>     print(f"{training_args.ipu_config_name = }")
+>     print(f"{training_args.ipu_config_name if training_args.ipu_config_name else model_args.model_name_or_path = }")
+> 
+298,299d311
+<     logger.setLevel(logging.INFO if is_main_process(training_args.local_rank) else logging.WARN)
+< 
+301,304d312
+<     logger.warning(
+<         f"Process rank: {training_args.local_rank}, device: {training_args.device}, n_gpu: {training_args.n_gpu}"
+<         f"distributed training: {bool(training_args.local_rank != -1)}, 16-bits training: {training_args.fp16}"
+<     )
+307,311d314
+<     # Set the verbosity to info of the Transformers logger (on main process only):
+<     if is_main_process(training_args.local_rank):
+<         transformers.utils.logging.set_verbosity_info()
+<     logger.info("Training/evaluation parameters %s", training_args)
+< 
+379a383,384
+>         if model_args.apply_spec_augment:
+>             raise ValueError("SpecAugment is not supported on IPU")
+381a387,392
+>     # IPU specific config updates
+>     config.update({"apply_spec_augment": False})
+> 
+>     # Whisper does not have a layer_norm_eps option, remains to be seen if this is a problem
+>     # config.update({"layer_norm_eps": 0.0001})
+> 
+401a413,417
+>     ipu_config = IPUConfig.from_pretrained(
+>         training_args.ipu_config_name if training_args.ipu_config_name else model_args.model_name_or_path,
+>         cache_dir=model_args.cache_dir,
+>         use_auth_token=True if model_args.use_auth_token else None,
+>     )
+446c462
+<     def prepare_dataset(batch):
+---
+>     def prepare_dataset(batch, feature_extractor, tokenizer):
+451a468,470
+> 
+>         igmin = inputs.get(model_input_name)[0]
+>         # print(igmin.shape)
+453c472
+<         batch[model_input_name] = inputs.get(model_input_name)[0]
+---
+>         batch[model_input_name] = igmin
+457a477,480
+>         if not training_args.fp32:
+>             # Cast audio inputs to FP16
+>             batch[model_input_name] = batch[model_input_name].astype(np.float16)
+> 
+463,469c486,491
+<     with training_args.main_process_first(desc="dataset map pre-processing"):
+<         vectorized_datasets = raw_datasets.map(
+<             prepare_dataset,
+<             remove_columns=next(iter(raw_datasets.values())).column_names,
+<             num_proc=data_args.preprocessing_num_workers,
+<             desc="preprocess train dataset",
+<         )
+---
+>     vectorized_datasets = raw_datasets.map(
+>         lambda batch: prepare_dataset(batch, feature_extractor, tokenizer),
+>         remove_columns=next(iter(raw_datasets.values())).column_names,
+>         num_proc=data_args.preprocessing_num_workers,
+>         desc="preprocess train dataset",
+>     )
+509,516c531,534
+<     # make sure all processes wait until data is saved
+<     with training_args.main_process_first():
+<         # only the main process saves them
+<         if is_main_process(training_args.local_rank):
+<             # save feature extractor, tokenizer and config
+<             feature_extractor.save_pretrained(training_args.output_dir)
+<             tokenizer.save_pretrained(training_args.output_dir)
+<             config.save_pretrained(training_args.output_dir)
+---
+>     # save feature extractor, tokenizer and config
+>     feature_extractor.save_pretrained(training_args.output_dir)
+>     tokenizer.save_pretrained(training_args.output_dir)
+>     config.save_pretrained(training_args.output_dir)
+518c536
+<     processor = AutoProcessor.from_pretrained(training_args.output_dir)
+---
+>     processor = WhisperProcessor(feature_extractor, tokenizer)
+524a543,545
+>         # pad_to_multiple_of=math.ceil(max_input_length),
+>         pad_to_multiple_of=80,
+>         pad_to_multiple_of_labels=training_args.generation_max_length,
+528c549
+<     trainer = Seq2SeqTrainer(
+---
+>     trainer = IPUSeq2SeqTrainer(
+529a551
+>         ipu_config=ipu_config,
+533d554
+<         tokenizer=feature_extractor,
+535a557,561
+>         eval_parallelize_kwargs={
+>             "use_cache": True,
+>             "use_cross_cache": True,
+>             "max_length": training_args.generation_max_length,
+>         },

From b42ddf8a19cae1073d6af1f7f54490d7df8e29da Mon Sep 17 00:00:00 2001
From: Callum McLean <callumm@graphcore.ai>
Date: Mon, 17 Jul 2023 14:51:27 +0100
Subject: [PATCH 4/8] `eval_parallelize_kwargs` ->
 `inference_parallelize_kwargs`

---
 examples/speech-recognition/run_speech_recognition_seq2seq.py | 2 +-
 1 file changed, 1 insertion(+), 1 deletion(-)

diff --git a/examples/speech-recognition/run_speech_recognition_seq2seq.py b/examples/speech-recognition/run_speech_recognition_seq2seq.py
index 9ba32c4ca..5509e1eee 100755
--- a/examples/speech-recognition/run_speech_recognition_seq2seq.py
+++ b/examples/speech-recognition/run_speech_recognition_seq2seq.py
@@ -548,7 +548,7 @@ def compute_metrics(pred):
         eval_dataset=vectorized_datasets["eval"] if training_args.do_eval else None,
         data_collator=data_collator,
         compute_metrics=compute_metrics if training_args.predict_with_generate else None,
-        eval_parallelize_kwargs={
+        inference_parallelize_kwargs={
             "use_cache": True,
             "use_cross_cache": True,
             "max_length": training_args.generation_max_length,

From c7c3f9eaa8ad6fbb331c9644a785449bea1659cd Mon Sep 17 00:00:00 2001
From: Callum McLean <callumm@graphcore.ai>
Date: Mon, 17 Jul 2023 15:26:53 +0100
Subject: [PATCH 5/8] Remove `pad_to_multiple_of`

---
 examples/speech-recognition/run_speech_recognition_seq2seq.py | 2 --
 1 file changed, 2 deletions(-)

diff --git a/examples/speech-recognition/run_speech_recognition_seq2seq.py b/examples/speech-recognition/run_speech_recognition_seq2seq.py
index 5509e1eee..b3fec1ae7 100755
--- a/examples/speech-recognition/run_speech_recognition_seq2seq.py
+++ b/examples/speech-recognition/run_speech_recognition_seq2seq.py
@@ -534,8 +534,6 @@ def compute_metrics(pred):
         processor=processor,
         decoder_start_token_id=model.config.decoder_start_token_id,
         forward_attention_mask=forward_attention_mask,
-        # pad_to_multiple_of=math.ceil(max_input_length),
-        pad_to_multiple_of=80,
         pad_to_multiple_of_labels=training_args.generation_max_length,
     )
 

From 53742af4c7da98a6ccc0513e20e87bc135673a9f Mon Sep 17 00:00:00 2001
From: Callum McLean <callumm@graphcore.ai>
Date: Mon, 17 Jul 2023 15:36:09 +0100
Subject: [PATCH 6/8] Redo diff file

---
 .../run_speech_recognition_seq2seq.txt        | 51 ++++++++-----------
 1 file changed, 21 insertions(+), 30 deletions(-)

diff --git a/tests/examples/run_speech_recognition_seq2seq.txt b/tests/examples/run_speech_recognition_seq2seq.txt
index 81d3cb0b9..d11b90719 100644
--- a/tests/examples/run_speech_recognition_seq2seq.txt
+++ b/tests/examples/run_speech_recognition_seq2seq.txt
@@ -44,61 +44,54 @@
 <         return batch
 ---
 >         return batch.data
-280a289,294
+280a289,291
 >     if training_args.gradient_checkpointing:
 >         raise ValueError("Gradient checkpointing not supported.")
 > 
->     print(f"{training_args.ipu_config_name = }")
->     print(f"{training_args.ipu_config_name if training_args.ipu_config_name else model_args.model_name_or_path = }")
-> 
-298,299d311
+298,299d308
 <     logger.setLevel(logging.INFO if is_main_process(training_args.local_rank) else logging.WARN)
 < 
-301,304d312
+301,304d309
 <     logger.warning(
 <         f"Process rank: {training_args.local_rank}, device: {training_args.device}, n_gpu: {training_args.n_gpu}"
 <         f"distributed training: {bool(training_args.local_rank != -1)}, 16-bits training: {training_args.fp16}"
 <     )
-307,311d314
+307,311d311
 <     # Set the verbosity to info of the Transformers logger (on main process only):
 <     if is_main_process(training_args.local_rank):
 <         transformers.utils.logging.set_verbosity_info()
 <     logger.info("Training/evaluation parameters %s", training_args)
 < 
-379a383,384
+379a380,381
 >         if model_args.apply_spec_augment:
 >             raise ValueError("SpecAugment is not supported on IPU")
-381a387,392
+381a384,389
 >     # IPU specific config updates
 >     config.update({"apply_spec_augment": False})
 > 
 >     # Whisper does not have a layer_norm_eps option, remains to be seen if this is a problem
 >     # config.update({"layer_norm_eps": 0.0001})
 > 
-401a413,417
+401a410,414
 >     ipu_config = IPUConfig.from_pretrained(
 >         training_args.ipu_config_name if training_args.ipu_config_name else model_args.model_name_or_path,
 >         cache_dir=model_args.cache_dir,
 >         use_auth_token=True if model_args.use_auth_token else None,
 >     )
-446c462
+446c459
 <     def prepare_dataset(batch):
 ---
 >     def prepare_dataset(batch, feature_extractor, tokenizer):
-451a468,470
-> 
->         igmin = inputs.get(model_input_name)[0]
->         # print(igmin.shape)
-453c472
-<         batch[model_input_name] = inputs.get(model_input_name)[0]
+452c465
+<         # process audio length
 ---
->         batch[model_input_name] = igmin
-457a477,480
+> 
+457a471,474
 >         if not training_args.fp32:
 >             # Cast audio inputs to FP16
 >             batch[model_input_name] = batch[model_input_name].astype(np.float16)
 > 
-463,469c486,491
+463,469c480,485
 <     with training_args.main_process_first(desc="dataset map pre-processing"):
 <         vectorized_datasets = raw_datasets.map(
 <             prepare_dataset,
@@ -113,7 +106,7 @@
 >         num_proc=data_args.preprocessing_num_workers,
 >         desc="preprocess train dataset",
 >     )
-509,516c531,534
+509,516c525,528
 <     # make sure all processes wait until data is saved
 <     with training_args.main_process_first():
 <         # only the main process saves them
@@ -127,24 +120,22 @@
 >     feature_extractor.save_pretrained(training_args.output_dir)
 >     tokenizer.save_pretrained(training_args.output_dir)
 >     config.save_pretrained(training_args.output_dir)
-518c536
+518c530
 <     processor = AutoProcessor.from_pretrained(training_args.output_dir)
 ---
 >     processor = WhisperProcessor(feature_extractor, tokenizer)
-524a543,545
->         # pad_to_multiple_of=math.ceil(max_input_length),
->         pad_to_multiple_of=80,
+524a537
 >         pad_to_multiple_of_labels=training_args.generation_max_length,
-528c549
+528c541
 <     trainer = Seq2SeqTrainer(
 ---
 >     trainer = IPUSeq2SeqTrainer(
-529a551
+529a543
 >         ipu_config=ipu_config,
-533d554
+533d546
 <         tokenizer=feature_extractor,
-535a557,561
->         eval_parallelize_kwargs={
+535a549,553
+>         inference_parallelize_kwargs={
 >             "use_cache": True,
 >             "use_cross_cache": True,
 >             "max_length": training_args.generation_max_length,

From ee5282cd1697511a518f9fce4c23f045eed6578b Mon Sep 17 00:00:00 2001
From: Callum McLean <callumm@graphcore.ai>
Date: Mon, 17 Jul 2023 17:12:25 +0100
Subject: [PATCH 7/8] Set `inference_parallelize_kwargs` in `IPUConfig`

---
 .../run_speech_recognition_seq2seq.py         | 10 +++---
 .../run_speech_recognition_seq2seq.txt        | 33 +++++++++----------
 2 files changed, 21 insertions(+), 22 deletions(-)

diff --git a/examples/speech-recognition/run_speech_recognition_seq2seq.py b/examples/speech-recognition/run_speech_recognition_seq2seq.py
index b3fec1ae7..244a4a201 100755
--- a/examples/speech-recognition/run_speech_recognition_seq2seq.py
+++ b/examples/speech-recognition/run_speech_recognition_seq2seq.py
@@ -411,6 +411,11 @@ def main():
         training_args.ipu_config_name if training_args.ipu_config_name else model_args.model_name_or_path,
         cache_dir=model_args.cache_dir,
         use_auth_token=True if model_args.use_auth_token else None,
+        inference_parallelize_kwargs={
+            "use_cache": True,
+            "use_cross_cache": True,
+            "max_length": training_args.generation_max_length,
+        },
     )
 
     if model.config.decoder_start_token_id is None:
@@ -546,11 +551,6 @@ def compute_metrics(pred):
         eval_dataset=vectorized_datasets["eval"] if training_args.do_eval else None,
         data_collator=data_collator,
         compute_metrics=compute_metrics if training_args.predict_with_generate else None,
-        inference_parallelize_kwargs={
-            "use_cache": True,
-            "use_cross_cache": True,
-            "max_length": training_args.generation_max_length,
-        },
     )
 
     # 12. Training
diff --git a/tests/examples/run_speech_recognition_seq2seq.txt b/tests/examples/run_speech_recognition_seq2seq.txt
index d11b90719..e259c424a 100644
--- a/tests/examples/run_speech_recognition_seq2seq.txt
+++ b/tests/examples/run_speech_recognition_seq2seq.txt
@@ -72,26 +72,31 @@
 >     # Whisper does not have a layer_norm_eps option, remains to be seen if this is a problem
 >     # config.update({"layer_norm_eps": 0.0001})
 > 
-401a410,414
+401a410,419
 >     ipu_config = IPUConfig.from_pretrained(
 >         training_args.ipu_config_name if training_args.ipu_config_name else model_args.model_name_or_path,
 >         cache_dir=model_args.cache_dir,
 >         use_auth_token=True if model_args.use_auth_token else None,
+>         inference_parallelize_kwargs={
+>             "use_cache": True,
+>             "use_cross_cache": True,
+>             "max_length": training_args.generation_max_length,
+>         },
 >     )
-446c459
+446c464
 <     def prepare_dataset(batch):
 ---
 >     def prepare_dataset(batch, feature_extractor, tokenizer):
-452c465
+452c470
 <         # process audio length
 ---
 > 
-457a471,474
+457a476,479
 >         if not training_args.fp32:
 >             # Cast audio inputs to FP16
 >             batch[model_input_name] = batch[model_input_name].astype(np.float16)
 > 
-463,469c480,485
+463,469c485,490
 <     with training_args.main_process_first(desc="dataset map pre-processing"):
 <         vectorized_datasets = raw_datasets.map(
 <             prepare_dataset,
@@ -106,7 +111,7 @@
 >         num_proc=data_args.preprocessing_num_workers,
 >         desc="preprocess train dataset",
 >     )
-509,516c525,528
+509,516c530,533
 <     # make sure all processes wait until data is saved
 <     with training_args.main_process_first():
 <         # only the main process saves them
@@ -120,23 +125,17 @@
 >     feature_extractor.save_pretrained(training_args.output_dir)
 >     tokenizer.save_pretrained(training_args.output_dir)
 >     config.save_pretrained(training_args.output_dir)
-518c530
+518c535
 <     processor = AutoProcessor.from_pretrained(training_args.output_dir)
 ---
 >     processor = WhisperProcessor(feature_extractor, tokenizer)
-524a537
+524a542
 >         pad_to_multiple_of_labels=training_args.generation_max_length,
-528c541
+528c546
 <     trainer = Seq2SeqTrainer(
 ---
 >     trainer = IPUSeq2SeqTrainer(
-529a543
+529a548
 >         ipu_config=ipu_config,
-533d546
+533d551
 <         tokenizer=feature_extractor,
-535a549,553
->         inference_parallelize_kwargs={
->             "use_cache": True,
->             "use_cross_cache": True,
->             "max_length": training_args.generation_max_length,
->         },

From fe466efe3e5d800f8a24968f353b2ea0bc0b4194 Mon Sep 17 00:00:00 2001
From: Callum McLean <callumm@graphcore.ai>
Date: Tue, 1 Aug 2023 11:22:22 +0100
Subject: [PATCH 8/8] Try manually removing transformers version bit from diff
 file

---
 tests/examples/run_speech_recognition_seq2seq.txt | 4 ----
 1 file changed, 4 deletions(-)

diff --git a/tests/examples/run_speech_recognition_seq2seq.txt b/tests/examples/run_speech_recognition_seq2seq.txt
index e259c424a..d56c58c0e 100644
--- a/tests/examples/run_speech_recognition_seq2seq.txt
+++ b/tests/examples/run_speech_recognition_seq2seq.txt
@@ -20,10 +20,6 @@
 > from optimum.graphcore import IPUConfig, IPUSeq2SeqTrainer
 > from optimum.graphcore import IPUSeq2SeqTrainingArguments as Seq2SeqTrainingArguments
 > 
-51c52
-< check_min_version("4.31.0.dev0")
----
-> check_min_version("4.29.0")
 238a240,242
 >     padding: Union[bool, str] = "longest"
 >     pad_to_multiple_of: Optional[int] = None