[Feature] Support Mixtral 8x7b (#263)

* support mixtral 8x7b

* update requirements

* update transformers requirements

* update datasets requirements

* remove openpyxl requirements

---------

Co-authored-by: LZHgrla <linzhihao@pjlab.org.cn>

requirements.txt

@@ -1,2 +1,3 @@
 -r requirements/runtime.txt
 -r requirements/deepspeed.txt
+-r requirements/modelscope.txt

requirements/modelscope.txt

@@ -0,0 +1 @@
+modelscope

requirements/runtime.txt

@@ -1,14 +1,12 @@
 bitsandbytes>=0.40.0
 datasets
 einops
-fsspec<=2023.6.0
 lagent>=0.1.2
-mmengine>=0.9.1
-modelscope
+mmengine>=0.10.1
 peft>=0.4.0
 scipy
 SentencePiece
 tiktoken
 torch
-transformers>=4.32.1,<=4.34.0
+transformers>=4.32.1,!=4.34.1,!=4.35.0,!=4.35.1,!=4.35.2
 transformers_stream_generator

setup.py

@@ -126,7 +126,10 @@ def gen_packages_items():
             parse_requirements('requirements.txt'),
             'deepspeed':
             parse_requirements('requirements/runtime.txt') +
-            parse_requirements('requirements/deepspeed.txt')
+            parse_requirements('requirements/deepspeed.txt'),
+            'modelscope':
+            parse_requirements('requirements/runtime.txt') +
+            parse_requirements('requirements/modelscope.txt'),
         },
         zip_safe=False,
         entry_points={'console_scripts': ['xtuner = xtuner:cli']})

xtuner/configs/mixtral/mixtral_8x7b_full_oasst1_internlm_template_e3.py

@@ -0,0 +1,163 @@
+# Copyright (c) OpenMMLab. All rights reserved.
+from datasets import load_dataset
+from mmengine.dataset import DefaultSampler
+from mmengine.hooks import (CheckpointHook, DistSamplerSeedHook, IterTimerHook,
+                            LoggerHook, ParamSchedulerHook)
+from mmengine.optim import AmpOptimWrapper, CosineAnnealingLR
+from torch.optim import AdamW
+from transformers import AutoModelForCausalLM, AutoTokenizer
+
+from xtuner.dataset import process_hf_dataset
+from xtuner.dataset.collate_fns import default_collate_fn
+from xtuner.dataset.map_fns import oasst1_map_fn, template_map_fn_factory
+from xtuner.engine import DatasetInfoHook, EvaluateChatHook, ThroughputHook
+from xtuner.model import SupervisedFinetune
+from xtuner.utils import PROMPT_TEMPLATE
+
+#######################################################################
+#                          PART 1  Settings                           #
+#######################################################################
+# Model
+pretrained_model_name_or_path = 'DiscoResearch/mixtral-7b-8expert'
+
+# Data
+data_path = 'timdettmers/openassistant-guanaco'
+prompt_template = PROMPT_TEMPLATE.internlm_chat
+max_length = 2048
+pack_to_max_length = True
+
+# Scheduler & Optimizer
+batch_size = 1  # per_device
+accumulative_counts = 16
+dataloader_num_workers = 0
+max_epochs = 3
+optim_type = AdamW
+lr = 2e-5
+betas = (0.9, 0.999)
+weight_decay = 0
+max_norm = 1  # grad clip
+
+# Evaluate the generation performance during the training
+evaluation_freq = 500
+SYSTEM = ''
+evaluation_inputs = [
+    '请给我介绍五个上海的景点', 'Please tell me five scenic spots in Shanghai'
+]
+
+#######################################################################
+#                      PART 2  Model & Tokenizer                      #
+#######################################################################
+tokenizer = dict(
+    type=AutoTokenizer.from_pretrained,
+    pretrained_model_name_or_path=pretrained_model_name_or_path,
+    trust_remote_code=True,
+    padding_side='right')
+
+model = dict(
+    type=SupervisedFinetune,
+    llm=dict(
+        type=AutoModelForCausalLM.from_pretrained,
+        pretrained_model_name_or_path=pretrained_model_name_or_path,
+        trust_remote_code=True))
+
+#######################################################################
+#                      PART 3  Dataset & Dataloader                   #
+#######################################################################
+train_dataset = dict(
+    type=process_hf_dataset,
+    dataset=dict(type=load_dataset, path=data_path),
+    tokenizer=tokenizer,
+    max_length=max_length,
+    dataset_map_fn=oasst1_map_fn,
+    template_map_fn=dict(
+        type=template_map_fn_factory, template=prompt_template),
+    remove_unused_columns=True,
+    shuffle_before_pack=True,
+    pack_to_max_length=pack_to_max_length)
+
+train_dataloader = dict(
+    batch_size=batch_size,
+    num_workers=dataloader_num_workers,
+    dataset=train_dataset,
+    sampler=dict(type=DefaultSampler, shuffle=True),
+    collate_fn=dict(type=default_collate_fn))
+
+#######################################################################
+#                    PART 4  Scheduler & Optimizer                    #
+#######################################################################
+# optimizer
+optim_wrapper = dict(
+    type=AmpOptimWrapper,
+    optimizer=dict(
+        type=optim_type, lr=lr, betas=betas, weight_decay=weight_decay),
+    clip_grad=dict(max_norm=max_norm, error_if_nonfinite=False),
+    accumulative_counts=accumulative_counts,
+    loss_scale='dynamic',
+    dtype='float16')
+
+# learning policy
+# More information: https://github.com/open-mmlab/mmengine/blob/main/docs/en/tutorials/param_scheduler.md  # noqa: E501
+param_scheduler = dict(
+    type=CosineAnnealingLR,
+    eta_min=0.0,
+    by_epoch=True,
+    T_max=max_epochs,
+    convert_to_iter_based=True)
+
+# train, val, test setting
+train_cfg = dict(by_epoch=True, max_epochs=max_epochs, val_interval=1)
+
+#######################################################################
+#                           PART 5  Runtime                           #
+#######################################################################
+# Log the dialogue periodically during the training process, optional
+custom_hooks = [
+    dict(type=DatasetInfoHook, tokenizer=tokenizer),
+    dict(
+        type=EvaluateChatHook,
+        tokenizer=tokenizer,
+        every_n_iters=evaluation_freq,
+        evaluation_inputs=evaluation_inputs,
+        system=SYSTEM,
+        prompt_template=prompt_template),
+    dict(type=ThroughputHook)
+]
+
+# configure default hooks
+default_hooks = dict(
+    # record the time of every iteration.
+    timer=dict(type=IterTimerHook),
+    # print log every 100 iterations.
+    logger=dict(type=LoggerHook, interval=10),
+    # enable the parameter scheduler.
+    param_scheduler=dict(type=ParamSchedulerHook),
+    # save checkpoint per epoch.
+    checkpoint=dict(type=CheckpointHook, interval=1),
+    # set sampler seed in distributed evrionment.
+    sampler_seed=dict(type=DistSamplerSeedHook),
+)
+
+# configure environment
+env_cfg = dict(
+    # whether to enable cudnn benchmark
+    cudnn_benchmark=False,
+    # set multi process parameters
+    mp_cfg=dict(mp_start_method='fork', opencv_num_threads=0),
+    # set distributed parameters
+    dist_cfg=dict(backend='nccl'),
+)
+
+# set visualizer
+visualizer = None
+
+# set log level
+log_level = 'INFO'
+
+# load from which checkpoint
+load_from = None
+
+# whether to resume training from the loaded checkpoint
+resume = False
+
+# Defaults to use random seed and disable `deterministic`
+randomness = dict(seed=None, deterministic=False)

xtuner/configs/mixtral/mixtral_8x7b_qlora_oasst1_internlm_template_e3.py

@@ -0,0 +1,182 @@
+# Copyright (c) OpenMMLab. All rights reserved.
+import torch
+from bitsandbytes.optim import PagedAdamW32bit
+from datasets import load_dataset
+from mmengine.dataset import DefaultSampler
+from mmengine.hooks import (CheckpointHook, DistSamplerSeedHook, IterTimerHook,
+                            LoggerHook, ParamSchedulerHook)
+from mmengine.optim import AmpOptimWrapper, CosineAnnealingLR
+from peft import LoraConfig
+from transformers import (AutoModelForCausalLM, AutoTokenizer,
+                          BitsAndBytesConfig)
+
+from xtuner.dataset import process_hf_dataset
+from xtuner.dataset.collate_fns import default_collate_fn
+from xtuner.dataset.map_fns import oasst1_map_fn, template_map_fn_factory
+from xtuner.engine import DatasetInfoHook, EvaluateChatHook
+from xtuner.model import SupervisedFinetune
+from xtuner.utils import PROMPT_TEMPLATE
+
+#######################################################################
+#                          PART 1  Settings                           #
+#######################################################################
+# Model
+pretrained_model_name_or_path = 'DiscoResearch/mixtral-7b-8expert'
+
+# Data
+data_path = 'timdettmers/openassistant-guanaco'
+prompt_template = PROMPT_TEMPLATE.internlm_chat
+max_length = 2048
+pack_to_max_length = True
+
+# Scheduler & Optimizer
+batch_size = 1  # per_device
+accumulative_counts = 16
+dataloader_num_workers = 0
+max_epochs = 3
+optim_type = PagedAdamW32bit
+lr = 2e-4
+betas = (0.9, 0.999)
+weight_decay = 0
+max_norm = 1  # grad clip
+
+# Evaluate the generation performance during the training
+evaluation_freq = 500
+SYSTEM = ''
+evaluation_inputs = [
+    '请给我介绍五个上海的景点', 'Please tell me five scenic spots in Shanghai'
+]
+
+#######################################################################
+#                      PART 2  Model & Tokenizer                      #
+#######################################################################
+tokenizer = dict(
+    type=AutoTokenizer.from_pretrained,
+    pretrained_model_name_or_path=pretrained_model_name_or_path,
+    trust_remote_code=True,
+    padding_side='right')
+
+model = dict(
+    type=SupervisedFinetune,
+    llm=dict(
+        type=AutoModelForCausalLM.from_pretrained,
+        pretrained_model_name_or_path=pretrained_model_name_or_path,
+        trust_remote_code=True,
+        torch_dtype=torch.float16,
+        quantization_config=dict(
+            type=BitsAndBytesConfig,
+            load_in_4bit=True,
+            load_in_8bit=False,
+            llm_int8_threshold=6.0,
+            llm_int8_has_fp16_weight=False,
+            bnb_4bit_compute_dtype=torch.float16,
+            bnb_4bit_use_double_quant=True,
+            bnb_4bit_quant_type='nf4')),
+    lora=dict(
+        type=LoraConfig,
+        r=64,
+        lora_alpha=16,
+        lora_dropout=0.1,
+        bias='none',
+        task_type='CAUSAL_LM'))
+
+#######################################################################
+#                      PART 3  Dataset & Dataloader                   #
+#######################################################################
+train_dataset = dict(
+    type=process_hf_dataset,
+    dataset=dict(type=load_dataset, path=data_path),
+    tokenizer=tokenizer,
+    max_length=max_length,
+    dataset_map_fn=oasst1_map_fn,
+    template_map_fn=dict(
+        type=template_map_fn_factory, template=prompt_template),
+    remove_unused_columns=True,
+    shuffle_before_pack=True,
+    pack_to_max_length=pack_to_max_length)
+
+train_dataloader = dict(
+    batch_size=batch_size,
+    num_workers=dataloader_num_workers,
+    dataset=train_dataset,
+    sampler=dict(type=DefaultSampler, shuffle=True),
+    collate_fn=dict(type=default_collate_fn))
+
+#######################################################################
+#                    PART 4  Scheduler & Optimizer                    #
+#######################################################################
+# optimizer
+optim_wrapper = dict(
+    type=AmpOptimWrapper,
+    optimizer=dict(
+        type=optim_type, lr=lr, betas=betas, weight_decay=weight_decay),
+    clip_grad=dict(max_norm=max_norm, error_if_nonfinite=False),
+    accumulative_counts=accumulative_counts,
+    loss_scale='dynamic',
+    dtype='float16')
+
+# learning policy
+# More information: https://github.com/open-mmlab/mmengine/blob/main/docs/en/tutorials/param_scheduler.md  # noqa: E501
+param_scheduler = dict(
+    type=CosineAnnealingLR,
+    eta_min=0.0,
+    by_epoch=True,
+    T_max=max_epochs,
+    convert_to_iter_based=True)
+
+# train, val, test setting
+train_cfg = dict(by_epoch=True, max_epochs=max_epochs, val_interval=1)
+
+#######################################################################
+#                           PART 5  Runtime                           #
+#######################################################################
+# Log the dialogue periodically during the training process, optional
+custom_hooks = [
+    dict(type=DatasetInfoHook, tokenizer=tokenizer),
+    dict(
+        type=EvaluateChatHook,
+        tokenizer=tokenizer,
+        every_n_iters=evaluation_freq,
+        evaluation_inputs=evaluation_inputs,
+        system=SYSTEM,
+        prompt_template=prompt_template)
+]
+
+# configure default hooks
+default_hooks = dict(
+    # record the time of every iteration.
+    timer=dict(type=IterTimerHook),
+    # print log every 100 iterations.
+    logger=dict(type=LoggerHook, interval=10),
+    # enable the parameter scheduler.
+    param_scheduler=dict(type=ParamSchedulerHook),
+    # save checkpoint per epoch.
+    checkpoint=dict(type=CheckpointHook, interval=1),
+    # set sampler seed in distributed evrionment.
+    sampler_seed=dict(type=DistSamplerSeedHook),
+)
+
+# configure environment
+env_cfg = dict(
+    # whether to enable cudnn benchmark
+    cudnn_benchmark=False,
+    # set multi process parameters
+    mp_cfg=dict(mp_start_method='fork', opencv_num_threads=0),
+    # set distributed parameters
+    dist_cfg=dict(backend='nccl'),
+)
+
+# set visualizer
+visualizer = None
+
+# set log level
+log_level = 'INFO'
+
+# load from which checkpoint
+load_from = None
+
+# whether to resume training from the loaded checkpoint
+resume = False
+
+# Defaults to use random seed and disable `deterministic`
+randomness = dict(seed=None, deterministic=False)