src/lmms-eval/lmms_eval/models/simple/aero.py

import os
import warnings
from typing import List, Optional, Tuple, Union

import librosa
import numpy as np
import PIL
import torch
from accelerate import Accelerator, DistributedType
from accelerate.state import AcceleratorState
from tqdm import tqdm
from transformers import AutoModelForCausalLM, AutoProcessor

from lmms_eval import utils
from lmms_eval.api.instance import Instance
from lmms_eval.api.model import lmms
from lmms_eval.api.registry import register_model
from lmms_eval.models.model_utils.audio_processing import downsample_audio

warnings.filterwarnings("ignore")

from loguru import logger as eval_logger

DEFAULT_IMAGE_TOKEN = "<image>"
DEFAULT_VIDEO_TOKEN = "<video>"
DEFAULT_AUDIO_TOKEN = "<|AUDIO|>"


@register_model("aero")
class Aero(lmms):
    """

    Example usage:

    accelerate launch --num_processes 8 --main_process_port 30000 -m lmms_eval \
        --model aero \
        --model_args pretrained=$CKPT_PATH,attn_implementation="flash_attention_2" \
        --tasks $TASK \
        --batch_size 1 \
        --log_samples_suffix $TASK_SUFFIX \
        --output_path ./logs/ --verbosity DEBUG

    """

    def __init__(
        self,
        pretrained: str = "lmms-lab/Aero-1-Audio",
        revision: str = "main",
        device: str = "cuda",
        dtype: Optional[Union[str, torch.dtype]] = "auto",
        batch_size: int = 1,
        trust_remote_code: Optional[bool] = True,
        attn_implementation: Optional[str] = None,
        device_map: str = "",
        chat_template: Optional[str] = None,
        use_cache: bool = True,
        eos_token_id: int = 151645,
        pad_token_id: int = 151643,
        **kwargs,
    ) -> None:
        super().__init__()
        # Do not use kwargs for now
        assert kwargs == {}, f"Unexpected kwargs: {kwargs}"

        accelerator = Accelerator()
        if accelerator.num_processes > 1 and device_map == "":
            self._device = torch.device(f"cuda:{accelerator.local_process_index}")
            self.device_map = f"cuda:{accelerator.local_process_index}"
        else:
            self._device = torch.device(device)
            self.device_map = device_map
        if isinstance(dtype, str) and dtype != "auto":
            dtype = getattr(torch, dtype)

        self._model = AutoModelForCausalLM.from_pretrained(pretrained, revision=revision, torch_dtype=dtype, device_map=self.device_map, trust_remote_code=trust_remote_code, attn_implementation=attn_implementation)
        self.pretrained = pretrained
        self._processor = AutoProcessor.from_pretrained(pretrained, revision=revision, trust_remote_code=trust_remote_code)
        # Pad from left for batched generation: https://huggingface.co/docs/transformers/v4.39.3/en/model_doc/llava#usage-tips
        self._processor.tokenizer.padding_side = "left"
        self._tokenizer = self._processor.tokenizer
        self._config = self._model.config
        self.batch_size_per_gpu = int(batch_size)
        self.chat_template = chat_template
        self.use_cache = use_cache
        self.eos_token_id = eos_token_id
        self.pad_token_id = pad_token_id
        if accelerator.num_processes > 1 and device_map == "":
            assert accelerator.distributed_type in [DistributedType.FSDP, DistributedType.MULTI_GPU, DistributedType.DEEPSPEED], "Unsupported distributed type provided. Only DDP and FSDP are supported."
            # If you want to use DistributedType.DEEPSPEED, you have to run accelerate config before using the model
            # Also, you have to select zero stage 0 (equivalent to DDP) in order to make the prepare model works
            # I tried to set different parameters in the kwargs to let default zero 2 stage works, but it didn't work.
            if accelerator.distributed_type == DistributedType.DEEPSPEED:
                kwargs = {
                    "train_micro_batch_size_per_gpu": self.batch_size_per_gpu,
                    "train_batch_size": self.batch_size_per_gpu * accelerator.num_processes,
                }
                AcceleratorState().deepspeed_plugin.deepspeed_config_process(must_match=True, **kwargs)
                eval_logger.info("Detected that you are using DistributedType.DEEPSPEED. Make sure you run `accelerate config` and set zero stage to 0")
            if accelerator.distributed_type == DistributedType.FSDP or accelerator.distributed_type == DistributedType.DEEPSPEED:
                self._model = accelerator.prepare(self.model)
            else:
                self._model = accelerator.prepare_model(self.model, evaluation_mode=True)
            self.accelerator = accelerator
            if self.accelerator.is_local_main_process:
                eval_logger.info(f"Using {accelerator.num_processes} devices with data parallelism")
            self._rank = self.accelerator.local_process_index
            self._world_size = self.accelerator.num_processes
        elif accelerator.num_processes == 1 and device_map == "auto":
            eval_logger.info(f"Using {accelerator.num_processes} devices with pipeline parallelism")
            self._rank = 0
            self._word_size = 1
        else:
            eval_logger.info(f"Using single device: {self._device}")
            self.model.to(self._device)
            self._rank = 0
            self._word_size = 1
        self.accelerator = accelerator

    @property
    def config(self):
        # return the associated transformers.AutoConfig for the given pretrained model.
        return self._config

    @property
    def tokenizer(self):
        return self._tokenizer

    @property
    def model(self):
        # returns the model, unwrapping it if using Accelerate
        if hasattr(self, "accelerator"):
            return self.accelerator.unwrap_model(self._model)
        else:
            return self._model

    def split_audio(self, audio_arrays):
        CHUNK_LIM = 480000
        SAMPLE_RATE = 16000
        audio_splits = []
        # Split the loaded audio to 30s chunks and extend the messages content
        for i in range(
            0,
            len(audio_arrays),
            CHUNK_LIM,
        ):
            audio_splits.append(audio_arrays[i : i + CHUNK_LIM])
        return audio_splits

    @property
    def max_length(self):
        return self._max_length

    @property
    def batch_size(self):
        return self.batch_size_per_gpu

    @property
    def device(self):
        return self._device

    @property
    def rank(self):
        return self._rank

    @property
    def world_size(self):
        return self._world_size

    def tok_encode(self, string: str, left_truncate_len=None, add_special_tokens=None) -> List[int]:
        """ """
        add_special_tokens = False if add_special_tokens is None else add_special_tokens
        encoding = self.tokenizer.encode(string, add_special_tokens=add_special_tokens)
        # left-truncate the encoded context to be at most `left_truncate_len` tokens long
        if left_truncate_len:
            encoding = encoding[-left_truncate_len:]
        return encoding

    def tok_decode(self, tokens):
        return self.tokenizer.decode(tokens)

    def loglikelihood(self, requests: List[Instance]) -> List[Tuple[float, bool]]:
        raise NotImplementedError("TODO: Implement loglikelihood for Kino")

    def flatten(self, input):
        new_list = []
        for i in input:
            for j in i:
                new_list.append(j)
        return new_list

    def generate_until(self, requests: List[Instance]) -> List[str]:
        res = []

        def _collate(x):
            # the negative sign on len(toks) sorts descending - this has a few advantages:
            # - time estimates will always be over not underestimates, which is more useful for planning
            # - to know the size of a batch when going through the list, you know the first one is always the batch
            #   padded context length. this is useful to simplify the batching logic and more importantly to make
            #   automatic adaptive batches much much easier to implement
            # - any OOMs will happen right away rather than near the end
            toks = self.tok_encode(x[0])
            return -len(toks), x[0]

        # we group requests by their generation_kwargs,
        # so that we don't try to execute e.g. greedy sampling and temp=0.8 sampling
        # in the same batch.
        re_ords = utils.Collator([reg.args for reg in requests], _collate, grouping=True)
        chunks = re_ords.get_batched(n=self.batch_size, batch_fn=None)
        num_iters = len(requests) // self.batch_size if len(requests) % self.batch_size == 0 else len(requests) // self.batch_size + 1
        pbar = tqdm(total=num_iters, disable=(self.rank != 0), desc="Model Responding")
        for chunk in chunks:
            contexts, all_gen_kwargs, doc_to_visual, doc_id, task, split = zip(*chunk)
            task = task[0]
            split = split[0]
            batched_visuals = [doc_to_visual[0](self.task_dict[task][split][ids]) for ids in doc_id]
            flattened_visuals = self.flatten(batched_visuals)
            batched_messages = []
            audios = []
            for visuals in batched_visuals:
                messages = [{"role": "user", "content": []}]
                for visual in visuals:
                    if isinstance(visual, dict) and "array" in visual:
                        splited_video_audio = self.split_audio(downsample_audio(visual["array"], visual["sampling_rate"], self._processor.audio_processor.sampling_rate))
                        audios.extend(splited_video_audio)
                        for _ in range(len(splited_video_audio)):
                            messages[0]["content"].append({"type": "audio", "audio_url": "<placeholder>"})
                batched_messages.append(messages)
            # we assume all gen kwargs in the batch are the same
            # this is safe to assume because the `grouper` object ensures it.
            gen_kwargs = all_gen_kwargs[0]

            context = contexts[0]
            for batch_number, context in enumerate(contexts):
                batched_messages[batch_number][0]["content"].append({"type": "text", "text": context})

            text = self._processor.apply_chat_template(batched_messages, tokenize=False, add_generation_prompt=True)

            if self.accelerator.is_main_process and doc_id[0] % 100 == 0:
                eval_logger.debug(f"Prompt for doc ID {doc_id[0]}:\n\n{text}\n")

            if len(audios) == 0:
                audios = None

            inputs = self._processor(audios=audios, text=text, sampling_rate=self._processor.audio_processor.sampling_rate, return_tensors="pt", padding=True).to(self._device, self.model.dtype)
            if "max_new_tokens" not in gen_kwargs:
                gen_kwargs["max_new_tokens"] = 1024
            if "temperature" not in gen_kwargs:
                gen_kwargs["temperature"] = 0
            if "top_p" not in gen_kwargs:
                gen_kwargs["top_p"] = None
            if "num_beams" not in gen_kwargs:
                gen_kwargs["num_beams"] = 1
            do_sample = True if gen_kwargs["temperature"] > 0 else False
            try:
                cont = self.model.generate(
                    **inputs,
                    do_sample=do_sample,
                    temperature=gen_kwargs["temperature"] if do_sample else None,
                    top_p=gen_kwargs["top_p"],
                    num_beams=gen_kwargs["num_beams"],
                    max_new_tokens=gen_kwargs["max_new_tokens"],
                    use_cache=self.use_cache,
                    pad_token_id=self.pad_token_id,
                    eos_token_id=self.eos_token_id,
                )
                cont = cont[:, inputs["input_ids"].shape[-1] :]
            except Exception as e:
                eval_logger.error(f"Error {e} in generating")
                text_outputs = ""
                res.append(text_outputs)
                pbar.update(1)
                self.cache_hook.add_partial("generate_until", (context, gen_kwargs), text_outputs)
                continue
            text_outputs = self.tokenizer.batch_decode(cont, skip_special_tokens=True)
            if self.accelerator.is_main_process and doc_id[0] % 100 == 0:
                eval_logger.debug(f"Generated text for doc ID {doc_id[0]}:\n\n{text_outputs}\n")

            for output, context in zip(text_outputs, contexts):
                res.append(output)
                self.cache_hook.add_partial("generate_until", (context, gen_kwargs), output)
            pbar.update(1)
        # reorder this group of results back to original unsorted form
        res = re_ords.get_original(res)

        pbar.close()
        return res

    def generate_until_multi_round(self, requests) -> List[str]:
        raise NotImplementedError("TODO: Implement multi-round generation for LLaVAHF")